[DISCONTINUED] MATLAB Answers Wish-list #2 (and bug reports)
Walter Roberson
on 29 Jun 2012
This topic is for features you would like to see for the MATLAB Answers facility itself, and also for bug reports about the MATLAB Answers facility.
This topic is the follow on to the earlier Wish-list for MATLAB Answer sections. That topic grew large enough to become unwieldy; and Mathworks has made enough changes to make a number of the past points no longer of relevance. There was also a more limited purpose <http://uk.mathworks.com/matlabcentral/answers/216662-new-design-of-the-forum-grey-on-white-wish-list-3-bug-reports
I suggest one wish (or bug report) per answer, so that people can vote their wishes.
227 Comments
For easy navigation and for the historians,
Anyone with edit privileges should feel free to update this whenever needed.
I sure wish Answers pages would load faster. Navigating the Answers section and sorting forum posts is remarkably slow, especially considering it's only text. And with the recent redesign of the site, there isn't even very much text at that.
Renato, this particular topic is about changes to the MATLAB Answers forum software, not about MATLAB itself.
Matlab lacks a more robust object oriented architecture that is related to how the majority of the myriad of functions are not structured in a hierarchical model, but rather exist on their own, loosely coupled one to another, even when there is a clear relationship between them.
In my simples example above, I compare the "Document Object Model" in a MS Office application (ie., Excel) with MatLab.
As you can see, if I want to graph a simple function, and at the same time change its axes, labels, and title... first, I have to know upfront the exact names of the specific five functions that would provide me those functionalities; Secondly, as I use the 'plot' function (the 'parent' function) by typing its name in the command window, the interface does not prompt me to the dependent ('child') methods and properties associated with the parent function; only allows me to complete the spelling (<Tab>)
What happens in more structured applications such as MS Excel, as you well know, is that beyond helping complete the spelling, there are more strict parent/child relationships between their (VBA) commands (objects). In such environments, a user could assume that typing 'plot' (or other meaningful command) in the command window's prompt, placing a period after the object handle and WITHOUT even pressing Tab, a dropdown list would appear revealing all of the associated / dependent commands (properties and methods) for the particular object/function to be immediately available for use.
In other words, there is no easy way to know what other commands I have available to use to manipulate/modify the graph/plot/parent object being used.
i wish matlab answer forum was quick and responsive as TI e2e community.
Are you talking about the speed of the forum software, or about the time it takes before a volunteer replies? Your post over here suggests that it is the time it takes people to reply that is your concern.
You have been posting questions about Simulink, PWM, Buck converters, PID control, and the like. There do not happen to be many volunteers who work on those topics. That is not the fault of the forum software or of the MATLAB language design.
The "Flagged Content" page is broken. It shows items that have been flagged but there is no way to get to them. When you click "an answer" it should take you to the answer that had been flagged but instead it just reloads the "Flagged Content" page.
It appears, on further reflection, that the case is this:
A Comment or Answer is flagged. The Question is deleted without deleting the individual Comment or Answer. Then the record of the flagging might still be in the list.
You can determine whether the comment or answer is still there or not by hovering over the "a comment" or "an answer" link: if it shows answers/flagged at the end of the URL then it is gone; it if is there it will have answers/######-Title-Here where ##### is the question number.
So at least the behavior is unexpected. I would expect that if a post has been deleted, that it would also be deleted from the "flagged content" list.
It is not always broken. As best I have deduced, if someone flags something and then someone else deletes it, the "flagged" entry in the Activity Log does not get removed at output time.
The contributor's page is broken, and currently only displays a bunch of # and commas.
Fixed.
I think that it would be better to have the forum adding automatically a target attribute
target="_blank"
to links present in threads, because we often want to keep the question/comment under our eyes when we click on such links.
I like it the way it is now.
Too_many_windows = clutter;
I just put in my vote!
@Walter: I know, but when it seems that we almost never want to open a link in the same window (which is often the case on forums where people answer questions), I think that it legitimately raises the question of whether it would be a good thing to have "open in new tab" implemented as a feature of the forum (by automatically adding a simple target attribute).
In Windows I have to right click and select "Open in new tab". I'd like that to be the default. On most web sites if you leave that site and go to another totally different site, it will open in a new window or tab, and if you stay on the same web site, it will open in the existing tab. Answers doesn't follow that de facto standard for some reason.
Firefox can be configured to either open links as tabs or windows. Then when I want to keep the original I open the new link using mouse-2 click (I think it is middle-button -- command-click on my MacBook Pro). Mouse-1 click (control-click, left-click) gives the menu option that includes the choice of new window or new tab for the link.
I don't know if it's just the two totally different computers I've tried it on (Dell desktop and HP laptop) or if it's due to some Java "upgrade" but in the last few days whenever I try to type an Answer or Comment into an edit box, the "Preview" box covers up almost all of the Save button. Disabling it doesn't help. All I can see is a tiny sliver of the left side and top side of the button. I've tried two different browsers (Firefox and IE) and tried changing the zoom level of the displayed page (control-mouse wheel) and nothing helps - it's still covered up. The Cancel button is totally covered so I can't see that button at all anymore. I have Java 7 update 25 (the latest version according to http://www.java.com/en/download/manual.jsp)
Is anyone else experiencing this problem with Answers?
I came back in to Firefox, without clearing the cache, and the problem has disappeared. All is normal again. So I guess you fixed it somehow, at least for now.
Clearing the browser cache worked (on Chrome).
I have the same problem.
Please try clearing your browser cache and reloading an Answers page. Let me know if things are any better...
Others are experiencing this issue as well (see this Newsreader thread). I am looking into it. Thanks for reporting the problem, I will respond here with any updates.
I think that it would be very interesting to have the possibility to tag/flag/mark/attach/bookmark threads so we can keep track and follow them without posting an answer or a comment in the threads.
"Bookmarked" threads could appear in an additional (private?) tab in our profile, e.g. at the end of the current series of tabs:
/ Profile \/ Questions \/ Answers \/ Comments \/ Bookmarks \
Ideally, we could add a label (a string with same size limitation as threads titles) and have entries sorted by label, which would allow us to implement the classification scheme that we prefer (date, name, tags, main topic, etc).
At this point, I print relevant mature threads as PDF, which is not possible when threads are just starting; in such case, I leave a tab open on each relevant young thread, hoping that I will remember to look at them before I mess up with my browser and lose all my tabs, .. which is not ideal.
I understand. I use Firefox so that's not an issue with me. I set up the syncing (it takes entering a code at first) so now syncing bookmarks between home and work computers happens automatically - it's really a nice feature.
Just because I am working from/on several computers, and if I am sync-ing files, I find it more difficult to sync bookmarks (especially because some of these machines have restrictions on browser types). I might just jump on this occasion to learn how to sync them I guess, but I also think that this could be a feature of the forum in the sense that most of our forum activity and interests would be centralized here.
Why not just bookmark discussions in your browser, like I do? For example I have a folder called Tutorials (under my MATLAB folder), and in there I have bookmarked links to common tutorial discussions, like how to ask a question, how to upload files, how to format questions, how do I get help on homework, etc. You could have a folder "Discussions to follow" or whatever you want to call it. It seems certainly better than exporting to a PDF.
It would be interesting to be able to cast votes AGAINST Answers (or Questions?), as well as in favor of them :-)
I haven't decided whether this should affect rep points negatively, but it should at least help sink the Answer to the bottom of the display. It would be a good way for the community to collectively warn someone away from a bad approach (e.g., abuses of EVAL)
I'd like to buy an argument.
I like it!
Actually you missed it, but the feature exists already ;-p
Ah well, too bad. I'm all in favor of a negative forum ;)
I used to (pre-MathWorks-employment) just write "-1 vote". This was discouraged since we want a positive forum.
Instead now I politely write why an answer is not the best...
Apologies if this has been suggested. The list has grown long and difficult to search....
I think there should be a privilege (earned at 1000 reputation points or so) that allows contributors to transform an Answer into a Comment (and vice versa?)
All too often, newcomers to the site don't understand how things work and instead of responding to Answers with Comments, they respond with Answers of their own. It would be good for trusted contributors to have the power to fix this.
On my display this is currently directly over my duplicate suggestion.
I knew I couldn't have been the first... Oh well, I +1'ed it. :-)
@IA, I think it sorts by number of votes (high to low) and then date (oldest to newest). It looks like you up voted my answer and it is now higher in the list...
Same here. I always wondered how it sorted the items. Like how this one could be added at some arbitrary place in the middle of a big long list, rather than at the end.
+1 !
Voting for comments would be a great feature to implement at the same time, by the way.
I think that's a great idea, and I voted for it. And I've always thought that the wish list and the bug list should be separate, rather than combined into one giant list.
An optional chat room with a main channel and thread-specific channels or user defined channels would be awesome. The rational is that sometimes/often we would be able to pinpoint much better the needs from the beginning after a short chat, than when we refine the question statement iteratively through dozens of comments (and it doesn't always converge!). The quality of answers could be significantly improved.
For thread specific channels, the log of the discussion would be saved in the thread in a way that we can edit (e.g. remove irrelevant entries and edit remaining entries, so we keep only relevant content).
It isn't the easiest feature to implement though, and it would be difficult to prevent people from just chatting without saving answers if they can create user defined channels. A solution could be to limit the creation of user defined channels to Editors, and keep the main room as well as thread specific channels for the rest of us.
I agree with that and this is why I added "optional". I didn't explain it well, but the typical situation that I had in mind was: I see that the OP is present on a thread at a given time and I propose to switch to chat if I am in the mood for that, in order to refine the question more efficiently. This interaction is logged (it ends up e.g. in the body of the question) and if I cannot answer on the fly, I can always say that I will think about it and post an answer later, and close the chat.
I know some people like chat systems a lot, and find them to be quite useful.
Speaking entirely for myself: I do not happen to like to use chat systems for technical matters, or use them with people I do not know relatively well. I end up feeling "locked in" to the conversation window, with its slow progress. With posting systems, I can concentrate on the material solidly, flipping windows as I need to to check something in help documentation or to run an example in MATLAB... or to go down and pick up a snack in the middle of doing something.
A failing formatting of code:
Here the formatting fails, when the space between "BC" and ">=" is removed:
|find(BC >= 2) .* find(BC <= 36)|
Working: find(BC >= 2) .* find(BC <= 36)
Failing: find(BC>= 2) .* find(BC <= 36)
Well, when dealing with regexp in MATLAB, aspirin is more important than even "indexing" itself ;) Therefore, I have a toolbox full of them..
But I should explain this pattern. Even if you are familiar with the topic, it might be useful to someone else.
The pattern can be seen as
(?<=expr_1)\|(?<content>expr_2)\|(?=exp_3)
----------- ------------------ ---------
look-behind token look-forward
where
\|(?<content>expr_2)|\
defines that I want to match something that lies within pipes "|" (that have to be escaped because pipes have a special meaning otherwise). This "thing" will be characterized by the pattern expr_2 and I want to save it as a named token (which I chose to name "content") that I can reuse later in the replacement (see in the def. of rep). In the expression
.*?[^\s\|]
. is a wildcard for any single char, * means that it can be repeated between 0 and infty times, ? makes the * lazy, meaning find as little as possible, otherwise we have a greedy expression that will roughly take everything between the first and the last pipe of str. The [^expr] matches any character that is not listed in expr, which means that we won't match pipes as closing pipes if they are preceded by a whitespace \s or a pipe.
The first part
(?<=\s|^|\<.)
is a positive look-behind; (?<=expr_1) defines that whatever comes after the look-behind is a match only if it is preceded by something that is matched by the look-behind. Here, expr_1 matches whitespace \s, or the beginning of the string ^, or the beginning of a word \<..
The last part
(?=[\s;,\.])
is a positive look-forward; (?=expr_3) defines that whatever comes before the look-forward is a match only if it is followed by something that is matched by the look-forward. Here, expr_3 is a set [expr] of possible matches, that contains whitespace \s, any punctuation ":", ",", and "." (that has to be escaped because . have a special meaning otherwise).
@Cedric: I do not have the toolbox, which contains the aspirin command. However, '(?<=\s|^|\<.)\|(?<content>.*?[^\s\|])\|(?=[\s;,\.])' looks like something harder than aspirin, or like you have rolled an angry armadillo over your keyboard.
I played a bit just for the fun of using regexp+aspirin..
EDIT: last update 02/06/2013 @ 3:22pm EST
str = ['|A|, \t |A| \t |A|..\n', ...
' |A|, \t |A| \t |A| ..\n', ...
'|A B|, \t |A B| \t |A B|..\n', ...
'|A|B|, \t |A|B| \t |A|B|..\n', ...
'|A||B|, \t |A||B| \t |A||B|..\n', ...
'|A | B|, \t |A | B| \t |A | B|..\n', ...
'|A || B|,\t |A || B| \t |A || B|..\n', ...
'|()|, \t |()| \t |()|..\n', ...
'|[]|, \t |[]| \t |[]|..\n', ...
'|{}|, \t |{}| \t |{}|..\n', ...
'|32|, \t |32| \t |32|..\n', ...
'||, \t | \t |..\n', ...
'"||", \t "|" \t "|"..\n'] ;
pat = '(?<=\s|^|\<.)\|(?<content>.*?[^\s\|])\|(?=[\s;,\.])' ;
rep = '$<content>' ;
newStr = regexprep(str, pat, rep) ;
fprintf(newStr) ;
Output:
A, A A..
A, A A ..
A B, A B A B..
A|B, A|B A|B..
A||B, A||B A||B..
A | B, A | B A | B..
A || B, A || B A || B..
(), () ()..
[], [] []..
{}, {} {}..
32, 32 32..
||, | |..
"||", "|" "|"..
That's maybe why I never use monospacing formatting. It never seemed to work right. If I want monospaced text, I format it as code. I wish we had the ability to specify whatever font (color, size, font name, underlining, etc.) we wanted.
Another feature that I would really appreciate is a select instead of the checkbox "Email me when answers are added to this question", that would offer more options. For example:
- Email me when answers are added to this question
- Email me when answers or comments are added to this question
- Email me when there is any activity on this thread
With that (and because the 3rd option could generate a lot of emails), I would also be interested in being allowed to choose the frequency of the emails with a second select (or something in my profile options), e.g.
- Email instantly..
- Daily summary..
- Weekly summary..
The rational for me is: I don't ask that many questions, but when I do I am genuinely interested in any piece of information that I can gather from anything on the thread/page. I don't really make a distinction between comments and answers in fact, in the sense that I could often define The Answer as the outcome of multiple answers and comments. I am also interested in being warned if any answer or comment is edited, because if the person who answered/commented believes that it is worth bringing a modification, then the update is important to me. I would therefore set my default to be the last item of the list/select mentioned above.
I am game for receiving these updates as well; that would teach me English in addition to MATLAB! ;)
Though with respect to editing, the Community Editors often edit for format or to correct typos.
As mentioned here, I think that having the possibility for the OP and high rep. members to add a top entry in the thread (like an answer but with a different design, e.g. greenish background or some frame) called "Synthesis" or "Conclusion", that would contain a synthesis/conclusion of the thread, would be interesting. This should be optional and worth no reward, so we have only valuable information, but highly prioritized by the search engine. The name of contributors should also be recorded in the sense that, at least to me, they are often a search criterion as important as keywords.
You can find an example of what I have in mind here, where I edited the question with a top part that synthesizes the thread: http://www.mathworks.com/matlabcentral/answers/59491-is-it-possible-to-set-watchdogs-memory-exec-time-in-matlab
The idea is really to allow people who care for that to bring added value to their threads. This could be especially useful for complicated threads, where finally a person is able to draw a conclusion out of dozens of answers/comments. The conclusion is really important information, yet it doesn't appear in the thread ultimately.
As a side note, I think that having the option not to prop. the thread at the top of the list when we edit the synthesis could be useful.. in my case, for example, I am pretty sure that people will start being tired about my "watchdog" thread if it comes back at the top of the list each time I add a coma to the text.. ;)
I think that if people were reminded (e.g. weekly) on a verified email that they have to come back to "close" their threads (or to actively indicate that they could still use help), more people would at least come back and see what was answered/commented. I am saying that, because I really suspect that many people actually find some way to do what they want after posting the question, and never come back to check their threads. All of you folks who have been active here for a long time, have roughly a 1/3 ratio of accepted answers, but when I look at the threads it seems that a significant amount of the answers that were not accepted are just a consequence of people not coming back..
Associated with this mechanism, there could be a special type of threads for open questions, in some sense a "discussion thread", like this one or any of those that should not have a particular answer chosen to be The answer.
Cheers,
Cedric
I am trying to find unanswered questions with 0 votes, but I want them listed newest to oldest. If I search for unanswered and sort by votes, I get all the questions with 0 votes sorted by date, but oldest to newest, followed by all the questions wit 1 vote, again sorted oldest to newest.
I am trying to close unclear and inappropriate questions which are unanswered. Basically I am trying to work through the unanswered questions. If the question is reasonable, I vote for it. If the question is inappropriate or unclear I add a comment saying what is inappropriate or unclear and close it. This leads me to needing the odd sort order/search capability.
I guess the first question is, is this filtering useful?
@Daniel, thanks for doing this!
Let Randy know if you notice any trends or anything else noteworthy or that could be useful to you.
I've asked hundreds of times for code formatting, but I definitely did not want the questions to be deleted, the users to be scared or the editors to waste time for checking.
But to stay on topic: More powerful search options would support your useful project to cleanup old not answered questions. And I appreciate your work.
According to about closing means that if the OP does not edit the question within 10 days, the question gets deleted. My feeling is that when we add comments like "format your code" or "what is your question" or tag it "doit4me", we should close the question. This has two effects. First the question disappears from the front page and other won't waste time looking at the question. Second, if the OP doesn't deal with it in 10 days, it gets deleted. I think both might cut down on the noise.
I have decided that what I am saying when I vote for an unanswered question is: Hey other contributors this question is not rubbish. There is potentially enough information to answer it.
Does closing mean, that an admin should decide if the question is deleted? If so, an excessive closing of question would be less useful than spending the time to answer old questions, which have not been answered already.
Only 2242 of 51314 questions have been voted yet, what is a sad ratio in my opinion. Therefore more votes are something I appreciate. But I'm not sure if all not unclear questions without an answer deserve a vote. Anyhow, your idea is useful, I'd prefer the criterion "vote for everything which you, Daniel, is worth to be voted", and a more flexible sorting would be very important to handle such a large and growing database as the answers forum.
I've struggeled with the search in the FileExchang today: I've searched for submissions concerning Excel, but searching for excel or "excel" got 1604 hit, which mostly contain the term "excellent".
I wish to have a matlab answers website on my cell phone (a mobile version)
The RSS feed contains just a question, you can't read the comments and answers.
You can get an RSS feed of Answers.
I wish that questioner could have possibility to change his mind about his(her) acceptance decision. At least for a certain period.
New issue, I just noticed today. If I use the Link button to copy a link of a comment that is not one of the last 3 posted, then have that link be a link in another comment (referring reader to the first comment), OR even just paste that link into the address field (of two different browsers, IE and Firefox), it goes to the top of the posting, it does not scroll directly to the comment. An example of this is the link which is http://www.mathworks.com/matlabcentral/answers/42413#comment_102678. Some investigation leads me to think that the problem occurs when the comment referred to is not one of the last 3 comments posted, in other words, it's one of those that are normally hidden and you have to reveal by clicking "Show n older comments." Can this be fixed? Otherwise we can't link to "hidden" comments.
The #togglecomments apparent URL is javascript. The contents are already on your system, and are expanded or collapsed by your browser. But something changed and now those sections are treated more like internal HTML than like something that can be acted on.
By analogy: you cannot search HTML <- comment -> style comments using your browser find function, even though the HTML is there.
Well that I can understand because your browser only looks at what characters actually got returned from the web server. If they aren't there, it won't find them. But you'd think if you asked the web site server for a "hidden" URL, it should give it to you directly (all expanded), not send back the "parent" page with the actual URL hidden inside it and collapsed / not exposed.
I have seen it.
Related to this is that now my browser Find function can only find content that is expanded.
Can I at least get an acknowledgement that the Mathworks has seen and reproduced this bug or undesired feature?
It should be possible to change and save the default sort order.
I would like it if a person ended up on your profile (by clicking the link that is your name) that, in addition to the profile privileges, questions, answers, comments, and email, it also had a link to my File Exchange. If I didn't have anything in the File Exchange then the link wouldn't be there but it would be there if I did have something in the File Exchange. It makes sense that if someone is interested enough to look up a person's profile and all their answers, etc. that they might also have an interest in visiting the person's File Exchange to see what useful programs the person has uploaded. I'd also like to see my own. It would be convenient for me, if I want to recommend one of my File Exchange submissions to someone, that a link to my File Exchange were right there on the left side below My Questions, My Answers, My Comments. It would save me from having to bring up my bookmark toolbar, locate the bookmark to my File Exchange and click on it.
Now that we have a "My Comments" tab, I realize this is not what I wanted (although I appreciate TMW improving the interface based on our requests). I tend to use the My Questions and My Answers tabs to see if anything moved to the top. This way I know if something changed in a thread I am interested in. Since I rarely answer a question I ask, there is not too much redundancy.
Most of my comments are in threads that I have given an answer. This means there is a lot of redundancy between My Answers and My Comments. I end up looking at threads twice (once from the movement in My Answers and once from the movement in My Comments). I would like another tab of "My Threads" that lists all threads in which the user has asked a question, provided an answer, or made a comment. Ideally we could also "like" a thread which would add it to "My Threads". This would have to be different from voting since I vote on things that I don't want to follow.
Again, I am happy that we now have My Comments, but now I want even more ...
Thanks for the feedback—we are looking at better ways to keep track of questions that you're interested in.
Eliminate the "Answer Accepted" feature. Let reputation points be dictated entirely by votes.
The first problem with the "Answer Accepted" is that it puts the power to reward Answerers in the hands of the Questioners, where it doesn't belong. This incentivizes contributors to accomadate people looking for homework answers or to spoonfeed over-detailed solutions to people looking to do minimal work.
The second problem is, as others have mentioned, that a better Answer can always come along.
@Matt J: I do not see the contradiction. Although I'd prefer no direct correlation between points and privileges, I participate on the current system TMW has decided for. Obviously it works sufficiently at the moment.
I do not vote to provide a further step to gain privileges, but to mark useful answers and questions. An active and useful voting system need not be directly related to the highest privilege levels. But for the low privilege levels, a manual decision would be too time-costly with a comparable arbitrariness.
I'd personally prefer this less automatical method, because the Answers forum has no way to refuse editor powers
That seems to contradict your Comment from Oct. 3, Jan. If it would be better to take the power to award privileges out of the hands of non-administrators, then surely taking voter power away from Questioners would be a good step toward that, not to mention that it would cut down on doit4mes.
In other forums, the admins select the editors manually based on skill and sympathy. They consider suggestions by other editors. I'd personally prefer this less automatical method, because the Answers forum has no way to refuse editor powers, when somebody is obviously not able to handle them professionally.
It is not difficult to fake the counting system. A user, I do not know, had voted for 700 of my answers during one weekend. Although I had full powers already, this did not change my privilegs, but it was strange. Fortunately the admins detected the peak of votes and remove it manually, after there was no plausible explanation for this. This means, that inspite of the automatic system for gaining privilegs, there is still a manual and human control in the background. And therefore I trust the TMW team to care for the working capacity of this forum sufficiently: Even with a modified reputation system, the main goal of this forum is to help users to solve their Matlab problems.
Let reputation points be dictated entirely by Answer Accepted, and remove voting points
I don't see your reasoning there, Azzi. The idea of reputation points is to measure contribution to the site, to see who can be trusted with privileges. Why are the Questioners the only good judges of that? Why are they good judges at all? After all, they're not generally the experts. They're the ones coming to us with Questions...
Matt J, I would say exactly the opposite, Let reputation points be dictated entirely by Answer Accepted, and remove voting points, without removing the voting option.
By the way, answering some (relatively easy) questions was my way to learn more about Matlab programming tricks. It stimulates. I concider myself as advanced beginner.
@Daniel, or even worse - did they read the answer?
I find accepting answers to be very helpful. It lets me know if the OP is happy with the answer. It frustrates me when users don't accept an answer. I don't know if the OP is happy and has simply left or if they are waiting for a new way to solve the problem.
The majority of questions are technically doit4me, but most can be answered in five lines of code or less and are so trivial that they are unlikely homework problems. Every once and a while I get conned and solve what I later think is a homework problem.
While I would love to see more well thought out difficult questions, I like seeing answers for the non-homework doit4me questions (typically not tagged doit4me). There are so many tricks that can be played with things like bsxfun, ndgrid, regexp, that often you see multiple approaches to easy things.
@Matt J: I agree that the accepting answers feature and the homework questions interfere positively until a certain level. On the other hand an accepted answer is usually a reliable mark for a solved problem. This supports contributors who are here for replying, and users searching for an already found solution. These two points are more important than the side-effect of the homework cheaters.
Votes are rare in this forum, too rare in my opinion. I'm voting nearly any useful question and answer I read and perhaps I have distributed more reputation points than anybody else. If you do not like the positive interference of doit4me's and reputation points, I suggest, that you vote for good answers in these case: "What have you done so far"? As soon as we find a third co-worker, our 3*2 reputation points will top the 4 points of the accepted answer.
I would vote for answers, which explains the formatting of code. But unfortunately such instructions should appear in the comment section. What a pitty.
When I look at the top 50 reputation, there are only a quite small number whom I have any concern about whether they are deliberately going after "low hanging fruit".
You misunderstood a little about Answers: I currently have 9043 Answers, 2841 of which were Accepted. No-one else has more than 2907 Answers right now (735 Accepted) or more than 840 Accepted Answers (out of 2879 the person has given.)
Until a few months ago, there were no incremental privileges: the only level was 1500, at which point all privileges were granted. There is now a graduated system in which the final privilege level is 3000. In the time since then no-one has come even close to passing the old (1500) level, and everyone who was already past 1500 was "grandfathered" to keep all privileges. So next to nothing changed. A small number of people have crossed the new 500 privilege level, and in so doing have gained the ability to trigger an Accept on old Questions.
But with regards to my reaching the 1500 level and thereby gaining all privileges: in truth it was the other way around. When MATLAB Central decided to allow editors, they looked at my reputation at the time, rounded down, and used that as the trigger level. Thus, I was not aiming for anything, as at the time there was nothing to aim for :-)
Well, it could just be my imagination, but I think I see a greater preponderance of doit4me's here than in the Newsgroup and I think the leverage given to the Questioners is part of why.
You say you have more Accepted Answers than nearly anyone else, but might that be the reason you're more indifferent to doit4me's? You've already gained all the privileges available.
Hmmm. A potential problem, but not, it seems to me, a problem in practice: people would tend to "run out of gas" before getting all that far.
I have more Accepted Answers than nearly anyone else has Answers, but I actively refuse to doit4me on everything but short easy solutions. I may have a wee bit of a tendency to over-detail though ;-)
From my understanding no standard user with less than 1500 rep can have all privileges. Some TMW employees, however, have obviously been given extra powers.
If I look at my profile I see
Daniel is a MATLAB Answers editor, and has all privileges.
If I look at Santosh Kasula's profile (an employee of TMW) I see
Santosh Kasula is a MATLAB Answers administrator, and has all privileges.
This seems really useful for identifying TMW employees who have some extra rights associated with Answers.
I then look at Sean de Wolski's profile and see
Sean de Wolski is a MATLAB Answers editor, and has all privileges.
but of course Sean may have gained his powers based on rep alone so I went and looked at Andreas Goser, who currently has less than 1500 rep but works for TMW.
Andreas Goser is a MATLAB Answers editor, and has all privileges.
It seems like Andreas has been given powers because he works for TMW (which isn't a problem in my opinion). I expected, however, that he would have been classed as an administrator and not an editor.
What do the terms mean?
Administrators are MathWorkers who have internal privileges for maintaining the system. Generally, those of us who are administrators don't do a lot of answering, we just try to keep things running smoothly.
Editors are MathWorkers or community members who were granted privileges beyond what their reputation would normally earn. Generally, editors ask, answer, and maintain questions--providing real value compared with us administrators :-)
Another suggestion. I think every question deserves an answer (though, perhaps after some comments prodding the poster to clarify the question). Just like the "accepted answer" questions have the number of answers in a green font and box, I suggest that questions with 0 answers be in a red box with a red font. This might make them stand out more and get people to answer them.
I have to disagree with your premiss that "every question deserves an answer", at least when it comes to (Homework & No Attempt & Doit4me) questions. Current practice is to Close those as Inappropriate.
As you can with for the "Accepted", but it's green. Anyway, it's just my opinion that it would be simple to add (because they're already doing it for Accepted), follows/expands their existing paradigm, couldn't hurt, and might be useful.
I don't know. You can easily filter to only show unanswered question.
Perhaps there is a method of which I am not aware, but I would like to be able to track questions that I commented on and not actually answered.
This functionality has been added.
Searching for comments from one's Profile page might eliminate the ‘string-matched name’ problem, since that would ideally use one's Contributors number.
I like Ryan's ‘comment search’ idea! +1
thanks - will work I suppose. seems like it would be a simple feature to add.
Not ideal and will pick up other people with a stringmatched name:
commented_by:"ryan"
The close/reopen system needs a rework. I don't know how to see the reason that a question got closed. I also don't know if the OP can see the reason or if they get an alert. More importantly, if a question get closed simply editing the question should not be enough to reopen it. For example, in this question the edit was "P.S. Why is it (The thread) closed?". Edits to the question should raise a flag and confirmation from someone with enough rep to reopen the question.
The only way to deal with persistent reopening is to either give up or delete the question. Deleting a persistently reopened question will likely result in persistent reposting. While the ability to close a question is useful for cutting down spam, the current system is ineffectual.
It is already asked... but it isn't displayed.
Closing is the mechanism used when it is intended that the user be able to edit and reopen the Question. When it is intended to prevent that, the Question is Deleted instead.
Myself, I do not think that requiring the editors to vote on closing a Question is efficient enough in practice. We do not presently have a sufficient number of active trusted volunteers to sustain such a system. The volunteers are mostly here in pursuit of their technical interests, not to do administrative maintenance.
@Walter In an ideal world I think closing and reopening should be done by a voting system where to be eligible to reopen an edit had to be made. We close questions so infrequently that I don't think it matters. It is just frustrating when you close a question to see it reopened with no real change of content.
Perhaps "unclear" close status should allow reopen on edit, but "inappropriate" close status should require intervention ? Editors should be able to reopen without edit but lower reps should require edit ?
I agree, Daniel. While spams are deleted directly, closed questions have a chance to be updated. But the unclosing procedure is not clear.
In the Community Profile, edit your personal description and include a return in it so you have two distinct lines. This will be accepted. However, when your profile is displayed, the return will have been converted to <br/> which cannot then be gotten rid of except by not using a hard return.
The text editor can be improved. Beginners have some difficulties to deal with it. I don't like the preview in another window. It's more flexible if the window were we write the code is the same as the preview. I suggest when we click on the code button, a window or highlighted zone appears, then we are sur that our code is well formated
I would love code highlighting similar (okay, identical to) the MATLAB editor. I think it would encourage people to make use of the code tags more.
Also, if the code tags could be actual tags; [code][/code], I think it would be much more intuitive.
I want an Editor for Matlab Answers like the one on Facebook. On the top of your page you will have notifications related only to the questions you have answered
- number of news comments
- number of new answered questions
- number of new accepted aswers
and when you click on the notification you will be redirected to the question.
Why not to both
- your own Questions and
- Questions you have answered or have made comments on
A comment often is a kind of promise to return to the Question when OP has added more info
I still see the word "Disabl" in the preview, e.g. currently while I'm typing this answer. But at some other occurrences it has been improved to "Disable" already, e.g. in the preview of comments.
I still see "Disabl" sometimes.
It shows up as "Disable" in Answers for me.
[From Lucas' answer]
About accepting a 'comment' that is in an answer box. It has more to do with the interface, in my opinion. Anyone coming from any other place, (intuitively) you see a chat box, you write there (like most forums), but to actually comment, you have to hit "Comment on this Answer".
Possibly change the 'Answer' text box, to "Answer this question" button, then open the text box or something like that. Which might alleviate the answering a comment thing.
[Image Analyst's suggestion]
Maybe you'd want it to recognize that the person submitting an answer is actually the original questionner and to ask them if they really want to answer and accept their own question or comment on another person's answer? That might not be a bad idea.
I wish the answers software would automatically delete the tag: MATLAB.
Also "matlab code". And "help". And "1" and "123"...
I know what you mean, but it should possibly only do that if there are other tags. It is tricky for me to search for "no tags" in order to post the comment prompting people to tag properly.
Though I wouldn't mind if there were better tag prompting including a link to the tag guide...
Syntax Highlighting
In the meantime, I created a userscript that applies MATLAB syntax highlighting using google-code-prettify (it was originally intended for Stack Overflow). Project is hosted on GitHub.
That someone for TMW would chime in about the Wish list format (with or without the standard disclaimers about personal opinions ... ). Should we be sending bugs to files@mathworks.com, to TMW customer support, or posting them here. What about our suggestions for "product enhancements"? Is one long list better than multiple short lists? Is a list for bugs and a lists for wishes better? I think as a community we are pretty ambivalent about the format of our wish list, so if anyone at TMW who deals with this stuff has even a weak opinion it would be useful.
Has anyone noticed that the counts for the "Contributors" have stalled? No numbers have been updated for several days now. I've been stuck at 607 accepted answers even though I see answers being accepted. Likewise, everyone else's numbers have not budged in several days.
And I say again that the bugs and the wishes should be in two separate lists instead of one long massive combined list.
This should be fixed as of Sep 19, 2012. Sorry for the inconvenience.
@IA, I have reported this.
Or the bugs are fixed soon such that the corresponding messages can be deleted...
Daniel, Wish-list #2 was created because it became impractical to navigate the many many entries in the first wish-list. The possibility that #3 would be created at some point was mooted early on in #2, in the comments above, but there hasn't been enough response to that to determine criteria for the next split (which will happen, in time.)
@Lucas the point of creating a Wish-list #2 was the release of "Answers 2.0" which made many of the previous wishes obsolete. I think it is useful to keep all the wishes together.
The string ( R ) is displayed correctly in the code formatting:
isempty(R)
but too smart when displayed inside the bars | isempty ( R ) | : isempty®
[EDITED]: As Walter supposes: isempty™
One of the chat rooms we had here turned (C) into a coffee cup.
E.g. length(C) was length<coffee cup icon> a bug that quickly became a feature.
Likewise using TM as the argument gets you the trademark symbol.
Is there any protocol about bumping? A guilt free way to bump an unanswered, uncommented question would be nice. Perhaps have the option to bump only appear after a week has passed.
(Disclaimer: I have one unanswered question which I think might just be a 'No, not possible' or 'Upgrade your matlab'. It's an Eval question if anyone is interested.)
A polite bump is acceptable after a reasonable waiting time. Add a comment to the Question, or edit the original Question to add more information or clarify.
By popular demand, the forum should be completely converted over to the MATLAB Excel Answers forum. :=P
A proper set of moderator tools that allows for merging duplicate questions and moving answers to comments when appropriate.
Background: The S/N-ratio of Answer is low (IMO). Most answers solves someones acute problem, but deteriorates to noise over time. It is difficult to find relevant information in old threads. I have even failed to find answers, I believe that I have written. Google-searches return threads of Answer, etc.
My problem: I contribute a great deal to this noise. Often, I get a polite comment regarding my misunderstandings and I learn. Thanks to all! Recently, I learned about how subsref and subsasgn are used in matfile. I ought to go back a denoise some of my answers, but how. It happens I delete.
Proposal: Next to the BOLD and ITALIC buttons I would like a STRIKE-OUT button.
When there is a sufficiently long Tag already entered, and one requests to edit the tags, the long tag will extend past the right-hand boundary of the tag edit box.
When the tag edit box is not enabled, the display is better. I am not sure at the moment if it word-wraps or length wraps, but it does split into multiple lines.
Example tag (all as one single tag):
r2012a maple mupad symengine symbolic engine linsolve array arguments error
Expressions of the form (URL) with () immediately around and no < and > markup, are now being rendered in final output as (<URL>) but not being so rendered in preview.
For example if I enter (<http://www.mathwork.com>) which is '(' 'http://www.mathworks.com' ')' then the final output I get is the same as if I had entered (<http://www.mathworks.com>) with the < and > included.
This final output is not automatically detected as a link at the browser level (at least not Firefox). Most browsers automatically detect and internally convert URLs that have whitespace on both sides, such as http://www.mathworks.com which I have entered here with no markup at all but will appear as a link when rendered by most browsers.
Previously, (URL) was not transformed on output (and was usually not detected automatically by browsers.) Adding the < and > in the final output (but not in preview!) has no precedent here and doesn't help anything.
Also now transformed is (URL ) with no space between the '(' and the URL, but a space after the URL before the ')'. It is transformed to (<URL> ) with the < and > visually showing and a space before the ')'. This URL format is not detected automatically by browsers.
Remember, if you want an HTML link you need to translate to <A HREF="URL"> and not just to <URL>
I kind of wish that when we scrolled through the answers to questions (especially when there are many "answers" like this one), that this:
would follow you down the page.
To clarify my remark: if there is a fixed portion of a screen, and there is a vertical scrollable area, I have no difficulty. But last week I encountered a site which had an information box that scrolled along with the content, but as soon as you paused in the scrolling, the information box would scoot down to be in position again, so it was moving in the service of trying to stay still.
Yeah... I guess so. Fix frames would definitely help reduce nausea... maybe this could be a user preference that one would select in their profile (with no fixed frames as the default):
Use fixed frames to display answers? Yes <No>
I am okay with fixed frames beside scrolling areas, but boxes that "move down" to be always visible make me nauseous (in a literal sense.)
Some URLs created today are not rendering properly. <URL1 URL2> is rendering as <URL2> with the > in black and the rest in blue. Prior URLs would render as URL2 in blue with no < and no >
The URLs do render properly in Preview.
This is an idiosyncrasy that preserves backwards-compatibility with the publish syntax in MATLAB (if you create and publish a MATLAB file with the same markup you'll see the same behavior). We're considering changing the implementation so that <http://mathworks.com http://mathworks.com> and <http://mathworks.com MathWorks> are consistent.
When I ask a question, I see a very funny red splosh before "Question" and "Body". Is this a skeletonized dot, a fractal or a red fountain? If there is only one of them, I'd had checked my monitor for pixel errors of flyspeck.
I'm not convinced that the title line should be called "question". In fact, "question" is misleading here. And while I do not understand the reason, why this title needs at least 4 words, titles with the allowed 250 characters are a secure indicator of a bad question.
Jan, does the browser scaling you were doing explain this issue as well?
Looks like a red asterisk on my screen.
Yesterday, someone posted a Question with the title
Fourier-cosine transform (very important!)
After I removed the "(very important!") it refused to save because that was only 3 words. So I had to go in and add some noise words, making it "Fourier-cosine transform problem". Without repeating in the title the formula that the user wanted transformed, there wasn't much else to say. I guess the user could have asked "How to do a fourier-cosine transform of this formula?" but is that really any better?
I am getting the impression that 3 words is enough.
I will also add that since the interface mods, there have been a lot more people trying to put the entire question in the title.
San closed How to fix Non-constant expression or empty matrix
8 Jul 2012 at 15:40
That is, a poster with 0 Reputation was able to Close their own question ? I would not have guessed that would be possible.
(I have reopened the question since, as it had an interesting question.)
I am not sure of the utility of being able to Close your own Question? If it has no Answers and you feel it is Not Appropriate or Unclear, you could just delete it.
Are users able to Close their own Question after it has received an Answer? Then considering how difficult it is to people to locate and review closed questions, does this become a mechanism for people to pretty much Delete their Question, as long as no-one with sufficient reputation happens to go through the activity log and get curious about it within 10 days?
Yes, you can close your own questions. You can also delete your own questions (but not after they receive an answer).
For Questions with sufficiently long titles, none of the title will be rendered.
Noticed on the following title (which had 28 characters remaining)
I hav tried to implement marker controlled watershed segmentation algorithm but the trees are not delineated correctly.plz help me with code.i hav to count no.of trees in a setellite image.is there any other alg to do this
Top of the page, the gold-ish coloured title.
That evening I also noticed some cases where the title sometimes did not render for some short titles, but would eventually render when the page was refreshed. It could be, then, that if I had tried more times than I did, I would have seen the title appear. I did, however, cross-check between the .com and the .co.uk site.
Where is the title not rendering (in the edit field, at the top of the page, in the URL, etc.)?
How about a link on the left hand navigation for FAQ and posts like Walter's on tags?
I'll say it for you, "Where to upload your images"
And "How to ask a question" and "How to use markup"
There is a logical bug with accepting someone else's answer. The determination of whether that option appears does not depend upon when the answer was posted, only upon when the question was posted.
I see that the documentation says that it happens after 7 days of inactivity from the poster, which is consistent with the behavior we see, but my feeling is that after 7 days of inactivity would be better.
The deleted Question http://www.mathworks.com/matlabcentral/answers/43031 is not giving the Page Not Found error. Instead it is giving the left frame and only the "contact us" line to the right.
At the moment, 43031 is the newest Question, and it was deleted (by the poster) before a newer Question was posted, and (at the moment) no newer Questions have been posted.
The Page Not Found is given for other appropriate cases.
The situation exists on both .com and .co.uk
Editing of (your own?) Answer is not recognized as Activity for the purposes of listing the latest activity against each Question, nor for the purposes of determining the age sorting order for each Question.
Editing of an Answer is recorded in the Recent Activity list, though.
There are certainly some self-edits that are grammar or typo fixes. When, though, an Editor edits someone else's Answer, it is likely to be a reformat that might seriously affect people's ability to read the material.
The edits I do in the new Tags question are a mix: sometimes I'm just fixing up the text spacing, and sometimes I am adding new material. Likewise, edits in the How To Ask A Question topic are likely to be worth bringing to people's attention.
"Per", for one, adds a response and later comes back and edits significant new material into place. A few of the other posters have taking to using edits as a strategy.
Perhaps to some extent this matter is less serious now that Comments have full markup. Well, except for the "resource" topics where using Comments is not quite appropriate.
This is intentional--edits to answers and comments do not affect the latest activity. This is based on the premise that edits to answers and comments are more likely to be grammar or typo fixes, while edits to questions are more likely to warrant new attention.
Another possible way to handle this is a "minor edit" checkbox, like the one in many wiki programs. Minor edits would not affect latest activity, other edits would. Thoughts?
The infrequently-used "Close" button is uncomfortably positioned between the more commonly used "Edit" and "Delete" buttons. I know there is conformation required for Close, but it's still an accident waiting to happen and slows down use while people double-check they are in the right position.
The recent activity list records people adding tags, but not people deleting tags.
The Recent Activity list does not show any information about questions that have been deleted: previously shown information is removed.
I can see how this could happen if the data was being dynamically constructed from a database rather than being extracted from a log, but I'm a little uncomfortable with the resulting disappearance of information.
Deleting of Questions also does not appear in the Recent Activity list.
Pixel graphics are expanded by some percent and smoothed, such that even screenshots look blurred. See above. They look much better in their original resolution.
[SOLVED] This is caused by zooming using Ctrl-+ in the browser.
I got it: The fontsize of the forum is too small for my monitor resolution. Therefore I've expanded the view by Ctrl-+ in my Firefox browser. This increases the fontsize and the dimensions of some GUI elements, e.g. the useless gray border vanishes. But unfortunately this scales the pictures also.
The image looks the same to me on the forum as when I look at it directly.
We don't do any (intentional) server-side or client-side manipulation of images. On my computer the image above is identical to the source image on imgur.
Remind me of your browser/operating system combination, please, and I'll see if I can replicate the blurriness.
Editors need a way to locate Closed postings.
I've been wondering whether the automatic deletion of a previously-closed Question gets indicated in the activity log? I'm presuming that a re-open would get listed as an event.
Editors (at least) could use a way to see the reason that was given for the closure.
(Sean closed a Question a moment ago, right as I was contemplating what wording I would use to close it, so I wanted to see how he phrased it.)
It is not the HTML, but the JavaScript, which let us suffer. Waht about a new label for computers: Ready for Windows 9, DirectX 13 and MATLAB Answers 2.0?
The working area given to construct a comment (the built-in editor) is not very big. It is okay for short comments, but I find that in practice it ends up being common to post code in comments, code demonstrating how to apply the concepts of the answer or the comment stream, and in that situation the small size gets to be uncomfortable. I start looking for a drag handle to increase the size of the area...
I am seeing the drag handle in the systems I using now. I cannot, however, increase more than another 5 or perhaps 6 lines.
At least in my FireFox I do have such a drag handle in the comment input boxes. But without doubt a bigger default size would be really useful: Currently I increase the size manually every time I post a comment of more than 3 lines. While white space is lost space when reading, it is free space when writing. (A little bit exaggerated, I know)
This Question is already large enough to be notably slow when I access from home :( A faster computer would likely help, but IMHO, people should not end up pushed into buying a faster system just to deal with HTML bloat.
Now that the Link Edit Flag Delete buttons are along the right hand side for Comments, I find myself looking in the same relative location for Answers.
When I look at the space needed at the top of Answers, I see one line to identify the originator and date, one line to identify the editor and date, and space for "Accept this Answer". This aggregate space in turn is almost exactly the same vertical space as is required for the profile picture. If the four buttons were moved to be linear like they are for Comments, then as long as something reasonable could be found for locating space for the "Votes", the it looks to me like the Answer and Comments space could be widened out to the width of the entire inner frame.
Perhaps when bugs/features are addressed, they could be deleted (saving space) and a corresponding post added to the changes thread
Thanks for creating Wish-list #2! We will continue to monitor and leverage the feedback and suggestions as we plan future releases!
There should be a grace period of a few minutes before something shows up as edited. For example, if you answer a question and then immediately fix a typo, it shouldn't show up as 'edited'.
And a pictographic example:
Btw. why is this picture displayed so blurry? It seems like pixel-graphics are expanded by 2% before showing.
Thanks Sean. Most wiki software has a "minor edit" checkbox for things like typos. Perhaps a similar option would help here...
One-line questions, answers or comments waste too much space, while the unused borders are lost.
Write
more
slowly
:)
As explained in the wish-list #1 already, the space on top of a section is larger than the one on the bottom:
This block belongs to the sentence on top. It could be code also.
Now a new section starts, but the indent block seem to belong to this new section, because it is nearer. Because larger horizontal skipping is possible with HTML tricks only, it is hard to structure the text optically.
Solution: While I assume, that the spacing should be smaller on the top than on the bottom in most cases, making them equal might be the best compromise.
Another example:
This is line 1:
a = rand(1, 1);
This is line 2:
b = rand(2, 2);
While this is the 3rd line:
c = rand(3, 3);
Thanks Jan, I'll make sure this gets fixed.
The tag system needs an upgrade. Ideally, when you enter a tag it should provide a brief overview of what the tag means. Low rep users should only be able to apply existing tags. Moderate rep users should be able to create new tags (with a corresponding definition).
@IA: I assume the waste of space and the further reduction of information density is an intended bug. But others might claim it is a question of taste. Anyway, I agree and there should be a separate list of open bugs. This should be accompanied by a list of fixed bugs, or even better a version history of other changes also. And finally this list should be maintained by the admin team of TMW: they are the professionals and payed for maintaining the forum.
I think we should have a separate bug list, and feature enhancement (wish) list. That way we won't have to split them up so often. True, there are some gray areas where it's hard to tell if a behavior is just really annoying but operating as intended, or is in fact a bug. For example, when you want to bold or italicize a word and double click it to select it and then click the B or I icon. I guess you'd just have to make a judgment call in those cases.
Pictures should be hosted on a TMW server, because other services will delete the pictures after some time and then the corresponding thread becomes meaningless.
It would also be nice to be able to upload .mat files more easily, see also
Thanks for letting us know; we'll get to work on fixing it.
Randy, one problem that's bitten me several times. If you write a response, then click the paper clip icon to add a file, and then decide that you don't want to attach it then and click cancel, then it completely blows away the response you just entered! Thank goodness I have Lazarus installed ( http://getlazarus.com/download. It's saved me more times than I can count.
Great news!
This evening (2013 Sep 18) we released the ability for users to post images and files to our servers. Some additional details are available in this blog post.
Sometimes what I want is a way to "import" a file (upload) into an edit field. Copying and pasting has different size limits for different operating systems (and different browsers), so I have ended up having to copy and paste in sections (sometimes a fair number of such cycles, for larger texts.)
Personally I feel that really this should be a browser feature. But unfortunately browsers don't seem to have it... :(
Interesting...
Start a Comment (does not apply to starting an Answer), and enter some text, and then Cancel the Comment. Then (before navigating off the page) click the same control to start a Comment in the same place. The partly-entered text will re-appear.
I did not expect this behavior, but I think my vote would be to retain it.
(I think I understand what is causing the behavior, as I had earlier looked at the HTML source of some Answers pages.)
I had a case today where I was editing a Question to format it, realized that I had accidently deleted parts of the material, Canceled the edit, clicked to edit again... and it brought back the munged version rather than the original.
Fortunately the part I had inadvertently deleted in my (unsaved) editing turned out to only be a single character. If necessary, I would have gone up to my browser toolbar and requested the page be refreshed, but I don't know that people would tend to think of that.
So the behavior is of mixed blessings.
This is intentional behavior.
If one inserts a sufficiently long set of Tags in the tag editor (total number of characters being important rather than the number of tags), then eventually one of the tags will wrap down and move to the second line. One can tell that the tag will wrap because the current tag one is entering will stop being given in full, and will start scrolling in the area from the end of the previous tag.
Once one has a tag at the beginning of the second line, the number of characters that one can enter in the tag after that before this same kind of mini-scrolling occurs, is not determined by the available space to the end of the second line: instead the limit is at most the number of characters that were in the tag that wrapped down (and appears at the beginning of the line). For example if that tag happened to be 9 characters long, then the mini-scrolling would start after the 9th character of the tag one typed in afterwards.
In most of my tests, the limit was at most 1 character less than the previous tag, and occasionally even fewer were permitted; I only observed the maximum length once.
The field-scrolling at the end of the first line turns out to be a problem in itself. If the previously entered tags happen to take up most of the first line, then one can end up with just one or two characters available to enter the new tag in, with the entered tag being impossible to read during entry due to the scrolling in the narrow field.
I think this has gotten worse, not better. If there are lots of tags and you try to add a new one, you can only see a fraction of one letter as you type. You can't fully see what you typed until you type a comma or type the enter key. Has Randy noticed this bug? It seems like it just got worse about a couple of months ago.
There are also spacing oddities when editing existing tags.
Has someone already addressed the change from having our most recent answers listed first to (now) where our oldest answers are listed first when we click on our profile? I usually check this out to see if people had made comments to a question I answered, but (having answered ~180 questions, which is not a lot compared to the top contributors) I do not want to have to page through all my older answers to see if anything recent had updated.
I wish we had the ability to sort through my own old answers, too. Sometimes I want to view older answers before answering newer questions... so this could be helpful.
When I posted the above, I had forgotten about the commented_by: and answered_by: prefixes that now exist for searches.
@Walter: Google hits it first page if I write my name usually. Also, it indexes comments, which is nice.
Hi, Elige,
As of yesterday afternoon, My Answers is now "most recently modified first"
Sorting of your Answers would be useful.
Also useful would be to be able to search through your own Answers specifically ("I know I wrote about this before!"). And once that is possible, to narrow a more general search to a specific contributor or set of contributors ("I know Jan wrote about this before!")
Let's set a limit from the start, of the number of Answers before we move on to #3.
I was about to suggest "25", but as I've already entered 8, that might be a bit small. 50 Answers perhaps? Or seasonally? Or start a new one every time a new MATLAB version is released? ;-)
I imagine this one will eventually get long also. Maybe the bug list/wish list should be a wiki instead of a forum posting.
I think we should probably create a posting that discusses what Tags are and how and why to use them, and lists a number of the more common ones. I would include "doit4me" and similar content meta-tags, but I would also suggest that enough tags be included to allow people to see (for example) whether it is best to use "nn" or "ann" or "neural networks"
The Answers section has developed the ad-hoc standard of labelling some Questions as "tutorials", when those Questions are intended as step-by-step information about solving a particular task. There is also the off-site FAQ that gets used a fair bit (especially about numeric precision, file name loops, and variable name loops.)
I am finding that there is a category of often-repeated information that does not really fall into either of the above, and I am not certain how to deal with that category. An example of the category is my series of comments indicating why one should always uses specific handles for graphics operations, and showing how to do so. I refer someone back to it every week or two -- often enough that it should probably be extracted into something independent. But it isn't really a FAQ and it isn't really a tutorial. "Frequently Referenced Advice" ?
Some of Image Analyst's material probably falls into the same category, such as his color segmentation demo.
Suggestions on how to handle this?
I've been considering adding a "chapter" to the FAQ on image processing questions. For example I'd refer them to the Vision bibliography web site when they ask something like "How do I detect/recognize characters/license plates/cars/fruit/faces/fingerprints/edges/movement?" Or "How do I mask an image?" Or "How do I extract and save all frames from a video?"
The Link button is very convenient! But it is easy to end up clicking the link while it is being presented, which has the natural side effect of asking the browser to load that page... a bit of a nuisance as one is already there, and a Problem if the reason one was trying to copy the URL was that one was in the middle of composing a reply (bye-bye reply!)
I do not know if there is a solution to this that does not sacrifice functionality.
For consistency, Questions themselves should have the Link button. Yes, you can go up to your browser bar and copy the URL, but now that you have the buttons, people are going to start relying on it and will think less about copying from the browser bar.
Before the June 28, 2012 Answers changes, if one was editing and one highlighted text and clicked on the Code button, then after the operation was done, the selection was removed.
As of those changes, the selection goes away for a small number of seconds while the reformat is done, but then the selection color comes back, highlighting a wider but shorter area, not covering all of the height of the original selection but covering wider (possibly even some areas that have no input.)
Selection bug might possibly have been fixed after I wrote the above.
Selection bug is not fixed, in my tests in Firefox. In case the solution required a reload, I quit the browser and re logged in, but that did not help.
I see the same symptoms as before.
This selection bug should be fixed. After highlighting a section of text and clicking a button, the full text should be highlighted.
Also, before the Answers change, one was left at the end of the selection after applying Code. Now, one is left at the beginning of the selection, which is annoying when one is starting at the top of some text and proceeding down reformatting parts of it.
Walter, thanks for reporting this. I will take a look at it.
This original wish-list is now so slow to load that Firefox and Safari now time-out the javascript 3 to 6 times for me.
Randy Souza indicates "We're aware of this issue and are investigating it."
We deployed a fix this evening (Aug 15 2012) that should eliminate the JavaScript warnings. Please post here or email me directly (<firstname>.<lastname>@mathworks.com) if you still see them.
Javascript timeouts seem to have been eliminated as of last night, at least within Firefox.
http://www.mathworks.com/matlabcentral/answers/1325-what-is-missing-from-matlab also times out when I load it from home (via Safari)
A moment ago, this list, #2, timed out for me from home (Firefox)
The HTML for each Question page naturally has to include the HTML to render the formatted version of each Question and Answer and Comment. The new version that allows everything to be edited is implemented in such a way that (unlike the old version) it does not fetch the raw block of text at the time of the edit request: instead, each Question and Answer and Comment now has an HTML "input" field that is pre-populated with the raw block of text.
This would significantly increase the size of each HTML page, perhaps 35-40% for typical Questions once one accounts for all the other page elements. But the Wish-list Question is big so the page overhead would be a relatively small fraction of the total page, so the total HTML for that page must have be getting on to 45-48% larger than it was, not even taking into account all the extra HTML and javascript brought in with the new version. The style sheets for each element must have gotten significantly more complicated as well, and with the sheer number of fields that would appear, browsers must be taking a fair while to do the rendering.
A sufficiently large and complicated HTML page will take time to render. The Wish-list was already slow, but the extra size and complication must have pushed it to the point that most people's systems are timing out on rendering it.
The time-stamps for the activities log and the flagged content log appear to be in GMT. This is inconsistent with the other time-stamps that the user sees and so can lead to some confusion.
Not a major problem ;-)
Tags
You can also select a web site from the following list
Americas
- América Latina (Español)
- Canada (English)
- United States (English)
Europe
- Belgium (English)
- Denmark (English)
- Deutschland (Deutsch)
- España (Español)
- Finland (English)
- France (Français)
- Ireland (English)
- Italia (Italiano)
- Luxembourg (English)
- Netherlands (English)
- Norway (English)
- Österreich (Deutsch)
- Portugal (English)
- Sweden (English)
- Switzerland
- United Kingdom(English)
Asia Pacific
- Australia (English)
- India (English)
- New Zealand (English)
- 中国
- 日本Japanese (日本語)
- 한국Korean (한국어)