Thread Subject:

Mathworks.com: Answers

Dear readers,

Are you satisfied with the new Answers pages?

I've tried to answer some questions but was impeded by the web interface. Most likely caused by the old hardware of my internet laptop (PentiumII 360MHz) a keystroke takes about 1 second to appear in the message box, if the answer has more than a few lines. I'm simply not able to bear this.
Even if I use an external editor to create the answer, the preview-box decreases the usability subsatantially. It would be more helpful if the update is triggered manually by hitting a button.

I'm really happy about the possibility to enlarge the text box for answers. In opposite to that the fixed size box for "Comment on this Answer" is midget.

I cannot find any explanation about how I can vote an answer. Of course I clicked on the small triangle by accident and gave an unintented vote. But if all of the other mentioned 100,000 MATLAB Central members are as silly as I am and learn how to use the GUI by random clicking, there will be a certain bias in the votes.

Nevertheless, the Answers pages are a very good idea and I assume they will grow up to a valuable source for finding solutions.

Thanks, Jan

On 11-01-25 06:15 PM, Jan Simon wrote:

> Are you satisfied with the new Answers pages?
>
> I've tried to answer some questions but was impeded by the web interface. Most
> likely caused by the old hardware of my internet laptop (PentiumII 360MHz) a
> keystroke takes about 1 second to appear in the message box, if the answer has
> more than a few lines. I'm simply not able to bear this.
> Even if I use an external editor to create the answer, the preview-box
> decreases the usability subsatantially. It would be more helpful if the update
> is triggered manually by hitting a button.

The latency of echoing characters does get quite slow with longer answers. It
has averaged well over 1 second per character on some of my longer answers. It
is distracting, and leads me to make typos.


> I'm really happy about the possibility to enlarge the text box for answers. In
> opposite to that the fixed size box for "Comment on this Answer" is midget.

You can type more lines; you just can't see them all at one time. You can
navigate in the space using your arrow keys.


> I cannot find any explanation about how I can vote an answer. Of course I
> clicked on the small triangle by accident and gave an unintented vote. But if
> all of the other mentioned 100,000 MATLAB Central members are as silly as I am
> and learn how to use the GUI by random clicking, there will be a certain bias
> in the votes.

Can't say... I found the mechanism almost immediately.


 > Nevertheless, the Answers pages are a very good idea and I assume they
 > will grow up to a valuable source for finding solutions.

They need better definition of their purpose, or else people will post to both
the Answers and the newsgroup (which is already happening.)

I sent some comments and suggestions in already about the voting mechanism.

I have not really figured out why they implemented this; the thought has
passed through my mind that it is "Keeping up with the Jones's" -- something
done because other vendors are doing it.

"Think two, count blue." <roberson@hushmail.com> wrote in message <ihnqf1$7k7$1@nrc-news.nrc.ca>...
> On 11-01-25 06:15 PM, Jan Simon wrote:
>
> > Are you satisfied with the new Answers pages?
> >
> > I've tried to answer some questions but was impeded by the web interface. Most
> > likely caused by the old hardware of my internet laptop (PentiumII 360MHz) a
> > keystroke takes about 1 second to appear in the message box, if the answer has
> > more than a few lines. I'm simply not able to bear this.
> > Even if I use an external editor to create the answer, the preview-box
> > decreases the usability subsatantially. It would be more helpful if the update
> > is triggered manually by hitting a button.
>
> The latency of echoing characters does get quite slow with longer answers. It
> has averaged well over 1 second per character on some of my longer answers. It
> is distracting, and leads me to make typos.


I find it best to stare at the wall as I type. Then go back and fix typos at the end!

Dear Think two, count blue,

Sorry, you confuse my standard header scheme. Is "dear roberson" better?

> The latency of echoing characters does get quite slow with longer answers. It
> has averaged well over 1 second per character on some of my longer answers. It
> is distracting, and leads me to make typos.

Not even making typos, but deleting the wrong characters when I try to fix them.

> > In opposite to that the fixed size box for "Comment on this Answer" is midget.
> You can type more lines;

Yes, I know. But navigating in this peephole is not easy.
Obviously the usability of the GUI has some space for further improvements. The possibility to include pictures is in every case a great advantage!

> I have not really figured out why they implemented this; the thought has
> passed through my mind that it is "Keeping up with the Jones's" -- something
> done because other vendors are doing it.

I'm discussing in the German goMatlab forum also:
  http://www.goMatlab.de
There are some differences in the GUI (markups for code, smilies, etc), and the possibility to attach files. The later is extremly helpful, because you can embed pictures in the questions directly. E.g.:
  http://www.gomatlab.de/videoqualitaet-ist-mies-t15660.html
goMatlab.de has no reputation system, but of course all users have a built-in reputation system, because they remember who spent time to formulate a question as clear as possible, who answered frequently and who asked confused questions and did not care about further enquiries. Therefore I cannot comprehend the reputation building and voting. I do not see a benefit neither in the Matlab community (questions of newcommers are answered also with great care) nor in the real life. At least an enormous reputation on the Mathworks pages means, that one has a lot of spare time and does not invest all energy in the real job...

Anyhow, even if Answers will be just CSSM2.0 with pictures, it is a step in the correct direction. And I see enough space for further development and an improved usability.

Kind regards, Jan

"Matt Fig" wrote in message <ihnqks$mr8$1@fred.mathworks.com>...
> "Think two, count blue." <roberson@hushmail.com> wrote in message <ihnqf1$7k7$1@nrc-news.nrc.ca>...
> > On 11-01-25 06:15 PM, Jan Simon wrote:
> >
> > > Are you satisfied with the new Answers pages?
> > >
> > > I've tried to answer some questions but was impeded by the web interface. Most
> > > likely caused by the old hardware of my internet laptop (PentiumII 360MHz) a
> > > keystroke takes about 1 second to appear in the message box, if the answer has
> > > more than a few lines. I'm simply not able to bear this.
> > > Even if I use an external editor to create the answer, the preview-box
> > > decreases the usability subsatantially. It would be more helpful if the update
> > > is triggered manually by hitting a button.
> >
> > The latency of echoing characters does get quite slow with longer answers. It
> > has averaged well over 1 second per character on some of my longer answers. It
> > is distracting, and leads me to make typos.
>
>
> I find it best to stare at the wall as I type. Then go back and fix typos at the end!

I find it somewhat redundant to the newsgroup. Already, I find people referencing question numbers (e.g. question #302), where I don't know how to find them (questions do not have a reference number in my browser)

As far as voting...what's the purpose? I can see the reason for an "accepted answer", but voting?

I remember a post from awhile ago about newsgroup activity. It would be interesting to see the impact of TMW answers on activity.

Dear Matt,

> I find it best to stare at the wall as I type. Then go back and fix typos at the end!

Do you mean:
Then [I drink a cup of coffee, do the email and say hello to the colleagues before I] go back and fix the typos at the end [of the day]!
Jan

On 25/01/11 8:20 PM, proecsm wrote:

> I find it somewhat redundant to the newsgroup. Already, I find people
> referencing question numbers (e.g. question #302), where I don't know
> how to find them (questions do not have a reference number in my browser)

That would be me, _usually_ referring to a duplicate post so that the
admins can clean up afterwards.

If you look at the URL for any particular Question, you will see that
the very beginning of the last component of it is a sequential number,
followed by a dash, followed by the title of the question, URL encoded.
I gather that these are used as file names and the number is added to
make them unique (as there could be empty titles or duplicate titles.)

When I am referencing a previous Question, especially for the admins, I
have made it a practice to put a link to the post in question, and to
put the number as the link text instead of bothering to quote the entire
title. The number is enough to disambiguate and ensure that we are
talking about the same Question.

On 25/01/11 8:16 PM, Jan Simon wrote:
> Dear Think two, count blue,
>
> Sorry, you confuse my standard header scheme. Is "dear roberson" better?

You could treat it as an expression sequence ;-)

> Not even making typos, but deleting the wrong characters when I try to
> fix them.

When it has autowrapped at a non-alphanumeric character such as a
mathematical operator, it is not always obvious that it did so instead
of wrapping because of a space there. More than once I have backspaced
at the beginning of a line intending to undo the folding, only to find
that instead I have deleted the character it automatically wrapped on.
It is a nuisance.


> goMatlab.de has no reputation system, but of course all users have a
> built-in reputation system, because they remember who spent time to
> formulate a question as clear as possible, who answered frequently and
> who asked confused questions and did not care about further enquiries.
> Therefore I cannot comprehend the reputation building and voting.

There are always a lot of new people asking questions, and it is in some
sense beneficial for them to have an indication of how trustable a
respondent is.

"Jan Simon" wrote in message <iho0ef$7e4$1@fred.mathworks.com>...
> Dear Matt,
>
> > I find it best to stare at the wall as I type. Then go back and fix typos at the end!
>
> Do you mean:
> Then [I drink a cup of coffee, do the email and say hello to the colleagues before I] go back and fix the typos at the end [of the day]!
> Jan

Yeah, something like that!
Actually, the lag is not always there. From where I sit the lag is really bad sometimes and not even noticeable other times. It is not always a function of the length of the post either.

Dear Matt,

> Actually, the lag is not always there. From where I sit the lag is really bad sometimes and not even noticeable other times. It is not always a function of the length of the post either.

This sounds, like the preview is created on the server and sent through the net, such that server load and connection speed influences the reaction time to keystrokes?! Then answering to a question is already a denial of service attack.
Let me modify my earlier estimation from "space for further improvements" to "suboptimal". If a more direct formulation is more suitable in this English speaking community: "bad idea".

I do like the simple tagging method used e.g. in goMatlab.de:
  [code]a = 12[/code]
  Let me [b]emphasize[/b] this.
  [url]http://www.goMatlab.de[/url]
There you can also insert graphical smilies, but this leads to ugly confusions if somebody types "A(:, :)", so I think this is not helpful. But include small diagrams for Simulink blocks is very useful.
An advantage of this simple text tags is that you know exactly what you delete if you hit the backspace key, because there are no invisible control characters. A drawback is of course the ambiguities e.g. in "function [code]=foo(input)", therefore I'd prefer the "<code>" syntax, although or because it equals HTML.
And if HTML tags are used for formatting accompanied by a tiny CSS definition, there is no need for a server side created preview, because the client computer can display this very easy.

I see the point of "Think blue, count two." (sorry, I forgot the trailing dot the last time): In some sense it is beneficial for newcommers to estimate the quality of an answer.
On the other hand I think my reputation in the CSSM newsgroup is not too ruinous, although Rune is still in fear to agree with me. But the quality of some of my answers would be much worse, if the typos and wrong ideas based on using mainly the ancient Matlab 6.5 were not corrected by Matt, James, Walter, Bruno, IA, Jos, John and the other cssm cracks. Finally the quality of an answer cannot be measured reliably by the reputation number of the person, but by the lack of corrections by readers with a high reputation. To include the undocumented features, I'm personally using the "negative Bruno-Yair index" (nBY): A solution is good, if Bruno and Yair do not post a better one. ;-)

I see the effects of ratings and download counts in the FEX: It is obvious that being one of the top-100 must have a magic attraction forcing an impressing high number of people submitting a lot of junk and getting a lot of computers to download it using different IPs. The rating system suffers from the inactive users: I estimate that 1 of 1000 downloads is rated. Rating and voting is not trustworthy, as long as you cannot obtain a representative number of opinions.

Kind regards, Jan

Dear readers,

Some other questions concerning the MATLAB Answers:
1. It seems like the answers are not displayed in the chronological order anymore. But of course the best answers are found in a development interactively by several users. So an answer can contain comments to other answers, because this is more suitable than inserting them in the comment section. See:
http://www.mathworks.com/matlabcentral/answers/417-how-do-i-use-randperm-to-produce-a-completely-new-sequence-no-numbers-in-same-place

2. The delay after keystrokes is ridiculous. While the good old CSSM looks like in the old 14.400 Baud modem times, Answers feels like 75 bits per second. To be exact: I've measured 3.4 sec yesterday and this is only 2.35 Baud. I'm appalled by such a low usability implemented by a high tech company. It is not friendly to users.

3. Currently Answers suffers from the same problems as CSSM:
a) Questions containing to few details to be answered are sometimes not maintained by the OPs and further inquiries of others remain unreplied. Finally some bold members post some guesses about potential answers, but this will not be helpful for anybody to store such questions permanently.
b) Questions like "how can I create variables from dynamically by strings" and "why doesn't find(0.308 == x) anything" will be asked again and again. And again.
Are there any ideas how to cope with these two phenomenons? Or will Answers contain just 12000 questions about 0.1+0.1+0.1==0.3 ?

4. It does not look like there is much interest in discussing the new service or do I just post in the wrong forum? Would it be better to send emails to Helen?

Kind regards, Jan

On 1/25/2011 6:20 PM, proecsm wrote:

>
> As far as voting...what's the purpose? I can see the reason for an "accepted answer", but voting?
>

I have not looked or used "Answers", but a "voting" feature, is part of the
new social internet groupy thingy that is sweeping the internet.

Many forums have voting on questions. I am no psychiatrist or anything like
that, but I guess the idea of voting is to make people feel better
when someone gives their answer a good vote. Like when we were little
kids and the school teacher gave us a 5 star, we felt good.

--Nasser

On Jan 29, 1:54 am, "Nasser M. Abbasi" <n...@12000.org> wrote:
> On 1/25/2011 6:20 PM, proecsm wrote:
>
>
>
> > As far as voting...what's the purpose?  I can see the reason for an "accepted answer", but voting?
>
> I have not looked or used "Answers", but a "voting" feature, is part of the
> new social internet groupy thingy that is sweeping the internet.
>
> Many forums have voting on questions. I am no psychiatrist or anything like
> that, but I guess the idea of voting is to make people feel better
> when someone gives their answer a good vote. Like when we were little
> kids and the school teacher gave us a 5 star, we felt good.
>
> --Nasser

-----------------------------------------------
Google has voting stars. I rarely use them. I don't plan on using
answers. I can only spread myself so thin. No time to use Facebook,
MySpace, Twitter, and no time to use answers either.

Dear ImageAnalyst,

> Google has voting stars. I rarely use them. I don't plan on using
> answers. I can only spread myself so thin.

You are one of the most active CSSM users.
I'd expect that if Answers works (will work, would work?) as expected, a lot of frequent beginner question are solved automatically and you can concentrate on the interesting problems.
But: Your missing interest in MATLAB Answers is a valuable answer to my question. Thanks!

> No time to use Facebook, MySpace, Twitter, and no time to use answers either.

I'm not surprised. Google Groups tells me, that you have posted 434 messages in CSSM this month.

Kind regards, Jan

On 11-01-28 03:42 PM, Jan Simon wrote:

> Some other questions concerning the MATLAB Answers:
> 1. It seems like the answers are not displayed in the chronological order
> anymore.

When an answer is marked as Accepted, it moves to be the first answer. All
other answers are displayed after that in chronological order of the creation
of the Answer. Within any one Answer, comments are displayed in chronological
order of their creation.

> 3. Currently Answers suffers from the same problems as CSSM:
> a) Questions containing to few details to be answered are sometimes not
> maintained by the OPs and further inquiries of others remain unreplied.
> Finally some bold members post some guesses about potential answers, but this
> will not be helpful for anybody to store such questions permanently.

There is some effort being put towards solving this, but as the discussion was
in email, I am not at liberty to say more.

> b) Questions like "how can I create variables from dynamically by strings" and
> "why doesn't find(0.308 == x) anything" will be asked again and again. And again.

One vendor bug/question system I used did some kind of analysis of the reports
and before allowing the submission to go through would say things like, "Have
you read the following bug report? Have you read the following technical
solution? Do you still wish to go ahead with the submission?" It *was* able to
find useful material sometimes. But I'm not sure it would be practical to
write an AI that was able to detect all the different manifestations of
inexperience with binary floating point representation. Though if a neural net
(or whatever) caught even 10% of those, it might be a worthwhile endeavour...

Dear Think two, count blue,

> > 3. Currently Answers suffers from the same problems as CSSM:
> > a) Questions containing to few details to be answered are sometimes not
> > maintained by the OPs and further inquiries of others remain unreplied.

> There is some effort being put towards solving this, but as the discussion
> was in email, I am not at liberty to say more.

Are you employed by MathWorks? Is a secrete discussion by email helpful for this problem?
 
> Though if a neural net (or whatever) caught even 10% of those, it might be a worthwhile endeavour...

You could let the users give the suggestions, they have a lot of experiences with that. And if a question can be answered by the same old standard replies or the question has too few details to be answered at all, the thread should be deleted after some days to reduce the noise. Simple, efficient, no false rejections.
Should the points gained in deleted threads vanish also? How is this handled today?

If Answers should be a valuable list of useful questions and answers, it is necessary to remove the "not useful" threads. Imagine Answers in 2 or 10 years: While the most important 200 questions are getting more and more points, because they solve problems again and again efficiently, the number of junk threads will grow without a limit, fill the server and confuse the users. The first impression of the quality of the Answers list will be proportional to the number of total reputation points divided by the number of threads -- as a very vage psychological approximation. With a lot of junk, the list smells junky.
Therefore I started to delete my comments, if the discussion shows, that they do not help to clear the OPs question. Is this a wanted behaviour?

Are asking people wanted to accept an answer?

Kind regards, Jan

On 11-02-01 04:46 PM, Jan Simon wrote:
> Dear Think two, count blue,
>
>> > 3. Currently Answers suffers from the same problems as CSSM:
>> > a) Questions containing to few details to be answered are sometimes not
>> > maintained by the OPs and further inquiries of others remain unreplied.
>
>> There is some effort being put towards solving this, but as the discussion
>> was in email, I am not at liberty to say more.
>
> Are you employed by MathWorks? Is a secrete discussion by email helpful for
> this problem?

I am listed at Mathworks as a Technical Contact for the organization I work
for, one of the people authorized to initiate bug reports and deal with the
Mathworks support and sales organizations on behalf of the organization. That
places me under an NDA (Non-Disclosure Agreement) for my communications with
Mathworks, no matter how off-hand they may have been phrased.

Dear Think two, count blue,

> Dear Think two, count blue wrote:
> I am listed at Mathworks as a Technical Contact for the organization I work
> for, one of the people authorized to initiate bug reports...
> That places me under an NDA (Non-Disclosure Agreement) for my
> communications with Mathworks, no matter how off-hand they may have been phrased.

Therefore I've started this thread: I publish my ideas and impressions at first, and send them to TMW afterwards, such that TMW is not bound to my NDA.
Most of all I hoped to inspirate others to improve my ideas and get inspired by better ones.
But of course my emails to TMW contain more private estimations also, e.g. concerning my personal feelings when using some details of Answers, and I do neither want to publish them or see them publish by somebody else. Criticism, even if it is justifiable, should not be released, if it will not support and lead to improvement.

Thanks for your answer, Think two, count blue. Your denoted, but not formulated, information are promissing - but I'm not sure, what they promise.

Kind regards, Jan

Thanks for sharing this! I'lll make sure to forward this to the engineering team.

Best, Helen

> One vendor bug/question system I used did some kind of analysis of the reports
> and before allowing the submission to go through would say things like, "Have
> you read the following bug report? Have you read the following technical
> solution? Do you still wish to go ahead with the submission?" It *was* able to
> find useful material sometimes. But I'm not sure it would be practical to
> write an AI that was able to detect all the different manifestations of
> inexperience with binary floating point representation. Though if a neural net
> (or whatever) caught even 10% of those, it might be a worthwhile endeavour...

We are reading these messages. This is the first of many planned releases for Answers. So far, we are getting just over 38 questions a day. This exceeds our wildest goals for a launch like this.

A unifying goal of this project is to build an architecture that encourages good behavior.

As a simple example, newsgroups often get low signal to noise in a given post because the architecture automatically quotes everything in the reply. People often do not remove extraneous quotes, so the amount of new material to old is quote low as a thread progresses. The architecture here does not suffer from that.

Often in a newsgroup, a correct answer will be given, but it is somewhere in the middle of the thread, and is not clearly designated. It is hard to find.

If someone asks an unclear question, there is often back and forth until the question is clarified. Here, people can keep editing their question until it is clear. Ideally, clarification comments will be deleted as they are responded to. People that come to the site weeks and years later will only see a well formed question, the accepted answer and then community endorsed answers in decreasing order.

Many of the comments above suggest further architectural changes. It is our policy to not promise nor comment on proposed timelines for features. However, please be assured we are listening very closely to these discussions and hope to delight with new features.

Dear Doug,

> It is our policy to not promise nor comment on proposed timelines for features. However, please be assured we are listening very closely to these discussions and hope to delight with new features.

Since some hours the update of the preview is triggered by a button press, and not after each leystroke anymore. Instead of promising or commenting, TMW implemented an important improvent. This *delights* me.

I can comprehend, why TMW does not comment future features in public. Now you, Doug, take the chance to explain the purpose and intentions of MATLAB Answers with more details. Thanks for this helpful support! If TMW shares its interesting and forward-looking intentions, it seems like some CSSMers are itching to fill this format with precious replies.

Kind regards, Jan

"Doug Hull" wrote:
> Many of the comments above suggest further architectural changes. It is our policy to not promise nor comment on proposed timelines for features. However, please be assured we are listening very closely to these discussions and hope to delight with new features.

I just accepted an answer by mistake. I wasn't asked to confirm, and I cannot undo it. It's not the right answer either. That is very annoying!

Dear Oliver,

> I just accepted an answer by mistake. I wasn't asked to confirm, and I cannot undo it. It's not the right answer either.

I'd email files@mathworks.com

JAn

"Oliver Woodford" wrote in message <iif0pj$cvv$1@fred.mathworks.com>...
> I just accepted an answer by mistake. I wasn't asked to confirm, and I cannot undo it. It's not the right answer either. That is very annoying!

I don't know if this is what happened to you, Oliver. But when I asked a question then later went to comment on somebody's answer, I noticed that the "Select Answer As Best" button is about 1 pixel below the comment box. I had to be very aware as I was commenting that one lazy click on the mouspad (using a laptop at the time) could select the wrong answer. The button should be moved IMO.

"Matt Fig" wrote:
> I don't know if this is what happened to you, Oliver. But when I asked a question then later went to comment on somebody's answer, I noticed that the "Select Answer As Best" button is about 1 pixel below the comment box. I had to be very aware as I was commenting that one lazy click on the mouspad (using a laptop at the time) could select the wrong answer. The button should be moved IMO.

Exactly, Matt. I was going to leave a comment, but was slightly inaccurate with my click.

Dear Doug,

> Since some hours the update of the preview is triggered by a button press, and not after each keystroke anymore.

Wow! This feature has been removed again?!?
I was stressed by the annoying latency, then I was happy that the simple solution of a manually triggered preview update was implemented, and after an hour this feature disappeared again.

Dissappointed, Jan

Hi Jan -

The way this feature works is that it automatically updates your preview in the beginning, but once you have a longer post going, it switches to the preview button. Is it possible in your previous message, you had cut and pasted a chunk of text that pushed the content over the auto-preview limit?

Our engineers are investigating ways to improve the performance. As Doug has mentioned, expect to see new features and updates in the upcoming weeks and months!

Helen

> > Since some hours the update of the preview is triggered by a button press, and not after each keystroke anymore.
>

Thanks, Matt and Oliver. Good points on the usability concern. I've reported this to engineering and we will look at improving the layout.

Helen

"Matt Fig" wrote in message <iif4qv$a20$1@fred.mathworks.com>...
> "Oliver Woodford" wrote in message <iif0pj$cvv$1@fred.mathworks.com>...
> > I just accepted an answer by mistake. I wasn't asked to confirm, and I cannot undo it. It's not the right answer either. That is very annoying!
>
> I don't know if this is what happened to you, Oliver. But when I asked a question then later went to comment on somebody's answer, I noticed that the "Select Answer As Best" button is about 1 pixel below the comment box. I had to be very aware as I was commenting that one lazy click on the mouspad (using a laptop at the time) could select the wrong answer. The button should be moved IMO.

Helen:
1 Pixel-1width difference:
The FEX has something similar with Update File Delete File beside each other. Maybe TMW could put a spacer in there too.

Yes, very good point. Thanks!

Helen

"Malcolm Lidierth" wrote in message news:iih876$bik$1@fred.mathworks.com...

Helen:
1 Pixel-1width difference:
The FEX has something similar with Update File Delete File beside each
other. Maybe TMW could put a spacer in there too.

"Helen Chen" wrote in message <iih3ub$t92$1@fred.mathworks.com>...
> Hi Jan -
>
> The way this feature works is that it automatically updates your preview in the beginning, but once you have a longer post going, it switches to the preview button. Is it possible in your previous message, you had cut and pasted a chunk of text that pushed the content over the auto-preview limit?

I like the auto-preview functionality, but the constant preview generation slows typing speed just a bit too much. If a checkbox could be added giving users the option to preview either automatically or only when “Preview” is clicked, users would feel more in control and less confined; the status of the checkbox might even be checked when Submit is clicked to gain insight into user psychology. This would prevent a one-feel fits all solution since the amount of lag that is acceptable is entirely subjective.

"Jan Simon" wrote in message <ihp12r$f9r$1@fred.mathworks.com>...
> And if HTML tags are used for formatting accompanied by a tiny CSS definition, there is no need for a server side created preview, because the client computer can display this very easy.

I like the idea of recognizing HTML tags since most users at this forum are probably already familiar with the basics. Double carriage returns between paragraphs are currently preserved like the <p> tag, but there is no way to implement a <br> tag, and typing an HTML tag directly renders as a link. Since there is a bug in the current linking system anyway, as I'll describe in the next paragraph, maybe links could be redone so that basic HTML tags can be implemented.

When the URL in a link contains one or more spaces, the content after the first space is interpreted as text to be displayed instead of as part of the URL. For example, <http://www.mathworks.com/matlabcentral/fileexchange/?term=optimization toolbox Search Results> displays “toolbox Search Results” instead of “Search Results” and links to http://www.mathworks.com/matlabcentral/fileexchange/?term=optimization instead of to the intended http://www.mathworks.com/matlabcentral/fileexchange/?term=optimization toolbox .

To address both of these concerns simultaneously, maybe users could click a button for adding links, which in turn would insert the HTML code into the text box with the URL and text to display intact. Other HTML commands could then be recognized as well, and users would not need to learn to tweak one type of shorthand for Wikipedia, another for MATLAB Answers, and yadda yadda yadda.

Regards,
George Evers
http://www.georgeevers.org

Ctrl+Z doesn’t undo changes at MATLAB Answers in the 64-bit version of IE8 on my machine, though it works at other sites such as Yahoo Mail. I reverted to the default settings just to be sure the problem wasn't caused by customization, but to no avail. I think this might be related to the auto-preview functionality since I don’t know what else would act on the text to impair normal Windows functionality; but it’s odd that the same problem doesn't occur in Opera. Is there anything about the auto previewing functionality that depends on a usually 32-bit technology such as Java or Flash, which are not shared by the 64-bit version of IE8?

Thanks George! All good suggestions that I will forward along to the engineers.

Helen

"George Evers" wrote in message <iiqdqr$fcq$1@fred.mathworks.com>...
> Ctrl+Z doesn’t undo changes at MATLAB Answers in the 64-bit version of IE8 on my machine, though it works at other sites such as Yahoo Mail. I reverted to the default settings just to be sure the problem wasn't caused by customization, but to no avail. I think this might be related to the auto-preview functionality since I don’t know what else would act on the text to impair normal Windows functionality; but it’s odd that the same problem doesn't occur in Opera. Is there anything about the auto previewing functionality that depends on a usually 32-bit technology such as Java or Flash, which are not shared by the 64-bit version of IE8?

On 07/02/11 9:15 PM, George Evers wrote:

> When the URL in a link contains one or more spaces, the content after
> the first space is interpreted as text to be displayed instead of as
> part of the URL.

Not a bug. Your understanding of URLs is wrong.

URIs cannot contain spaces. Special characters such as space must be
encoded. Spaces get encoded as %20

http://tools.ietf.org/html/rfc3986#appendix-A

query = *( pchar / "/" / "?" )

pchar = unreserved / pct-encoded / sub-delims / ":" / "@"

sub-delims = "!" / "$" / "&" / "'" / "(" / ")"
               / "*" / "+" / "," / ";" / "="

unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~"


Space is not ALPHA or DIGIT or any of the listed characters, so it is
not permitted in a pchar and is thus not permitted in a query. It must
thus be pct-encoded to pass through.

"Think blue, count two." <roberson@hushmail.com> wrote in message <kHf4p.66526$TZ4.60390@newsfe20.iad>...
> URIs cannot contain spaces. Special characters such as space must be
> encoded. Spaces get encoded as %20

Think blue,

I see your point, but when typing directly into the address bar or within the parentheses of an <a href> tag, the higher-level human space works fine since conversion to the lower-level "%20" is automatic when the link is clicked. For example, if "http://www.mathworks.com/matlabcentral/fileexchange/?term=optimization toolbox" is pasted into the address bar or inserted as a link in Word, it works just fine. While you're correct that the unexpected behavior isn't technically a bug, I think of "%20" as an internal conversion that should be implemented automatically rather than being of concern on the user side; though I admit this is largely a subjective and aesthetic complaint, it is consistent with the now common practice of automatically translating spaces in %20's.

Regards,
George

On 11-02-08 03:56 PM, George Evers wrote:

> I see your point, but when typing directly into the address bar or within the
> parentheses of an <a href> tag, the higher-level human space works fine since
> conversion to the lower-level "%20" is automatic when the link is clicked.

It depends on the browser.

> For
> example, if
> "http://www.mathworks.com/matlabcentral/fileexchange/?term=optimization
> toolbox" is pasted into the address bar or inserted as a link in Word, it
> works just fine.

In Thunderbird (Usenet Reader) the http: is recognized right up to the ", but
the internal spaces are squeezed out of the URL to be accessed.

"Doug Hull" <hull@mathworks.SPAMPROOFcom> wrote in message <iibv5r$q8v$1@fred.mathworks.com>...
> We are reading these messages. This is the first of many planned releases for Answers. So far, we are getting just over 38 questions a day. This exceeds our wildest goals for a launch like this.
>

I'm still curious to know why this is going to be better than stackoverflow?
http://stackoverflow.com/questions/tagged/matlab

Adrian

"Matt Fig" wrote in message <iif4qv$a20$1@fred.mathworks.com>...
> "Oliver Woodford" wrote in message <iif0pj$cvv$1@fred.mathworks.com>...
> > I just accepted an answer by mistake. I wasn't asked to confirm, and I cannot undo it. It's not the right answer either. That is very annoying!
>
> I don't know if this is what happened to you, Oliver. But when I asked a question then later went to comment on somebody's answer, I noticed that the "Select Answer As Best" button is about 1 pixel below the comment box. I had to be very aware as I was commenting that one lazy click on the mouspad (using a laptop at the time) could select the wrong answer. The button should be moved IMO.

Are you as annoyed as I am at the size of the box provided to type in a comment to an answer? I feel like I can barely fit ten words in there. Maybe it's small to encourage short comments?

"Think two, count blue." <roberson@hushmail.com> wrote in message <iisfi5$17g$1@nrc-news.nrc.ca>...
>
> In Thunderbird (Usenet Reader) the http: is recognized right up to the ", but
> the internal spaces are squeezed out of the URL to be accessed.

Think two,

A valid point, but spaces are recognized at least by Opera, Chrome, Firefox, and IE, which are the only browsers I use for testing purposes. So their recognition is becoming quite common.

Zooming back out to the big picture, it would be nice for tagging to be HTML compatible, which currently is not an option since the <> symbols are used to denote links; this incompatibility could be removed with an Insert Link button whereby the user could paste the URL in one cell and type text to be displayed in another for automatic conversion to a standard <a href> tag; HTML-savvy users could bypass the button, which would mainly exist for those who aren't familiar with HTML. But even Myspace users are at least somewhat familiar with HTML these days, and its incorporation into online posting would allow more creative control on the user side.

Regards,
George
http://www.georgeevers.org

On 10/02/11 12:14 AM, George Evers wrote:
> "Think two, count blue." <roberson@hushmail.com> wrote in message
> <iisfi5$17g$1@nrc-news.nrc.ca>...
>>
>> In Thunderbird (Usenet Reader) the http: is recognized right up to the
>> ", but the internal spaces are squeezed out of the URL to be accessed.
>
> Think two,
>
> A valid point, but spaces are recognized at least by Opera, Chrome,
> Firefox, and IE, which are the only browsers I use for testing purposes.
> So their recognition is becoming quite common.

The RFC I referenced specifically says,

"In some cases, extra whitespace (spaces, line-breaks, tabs, etc.) may
have to be added to break a long URI across lines. The whitespace
should be ignored when the URI is extracted."

and

"For robustness, software that accepts user-typed URI should attempt to
recognize and strip both delimiters and embedded whitespace."

The use of whitespace within URI is directly contrary to the design
considerations, http://tools.ietf.org/html/rfc3986#section-1.2.1

"Sam asks Kim for a location to get more information, so Kim writes the
URI for the research site on a napkin. Upon returning home, Sam takes
out the napkin and types the URI into a computer, which then retrieves
the information to which Kim referred."

If Kim use spaces within the URI then Sam cannot determine from the
writing on the napkin how many spaces are used and at which places,
making the transcription ambiguous.

"Paul" said
> Are you as annoyed as I am at the size of the box provided to type in a comment to an answer? I feel like I can barely fit ten words in there. Maybe it's small to encourage short comments?

No nefarious scheme behind the box size - we didn't think comments would not be used as much as they are. Given the use model that we are seeing we are currently working on improving the usability of the comment box, so you will probably be seeing changes to this area in upcoming releases.

Thanks for the feedback! Keep it coming folks as your input will help us improve this application works for you. :-)

Helen

"Think blue, count two." <roberson@hushmail.com> wrote in message <YIL4p.72$2v4.48@newsfe17.iad>...
> The RFC I referenced specifically says,
>
> "In some cases, extra whitespace (spaces, line-breaks, tabs, etc.) may
> have to be added to break a long URI across lines. The whitespace
> should be ignored when the URI is extracted."

Think abstractly,

If the direction of evolution is toward higher-level language, I would expect the number of spaces between words to max out at one. The problem posed in the RFC could be solved simply by treating concurrent whitespaces as a single whitespace like the human mind does when reading manual writing, but the need to wrap a long URL onto multiple lines does not logically justify the entire removal of all spaces. Regardless of the logic, I agree that the specifications of the RFC should be followed. Had the RFC specification been followed across the board, the URL in question would not have contained a space at all since I pasted it from the address bar after a search at The File Exchange; this was a rare example of a browser unexpectedly displaying a space in the address bar [1], which is presumably due to RFC 3986 not having been followed consistently.

[1] http://forums.dpreview.com/forums/readflat.asp?forum=1018&message=34904983&changemode=1

> The use of whitespace within URI is directly contrary to the design
> considerations, http://tools.ietf.org/html/rfc3986#section-1.2.1

That is a good point; yet despite it, browsers seem to be evolving in a different direction. Indicating a particular disdain for decisions made solely on the basis of respect, Einstein said, "Blind respect for authority is the greatest enemy of truth." While it is not so much truth we have debated here as direction, the point that authority is not necessarily correct, which implies that arguments should be evaluated on their own merit, is still equally valid. The important questions are: "Which direction would users prefer?", and "Can we safely go in that direction?" While the apparent consensus among browser designers is that browsers should be capable of interpreting the space, which seems to acknowledge that there are at least some users who prefer its incorporation, I agree that it is safer to follow the RFC. Any problem with the RFC would ideally be presented through the proper
channels to avoid discrepant behavior.

> "Sam asks Kim for a location to get more information, so Kim writes the
> URI for the research site on a napkin. Upon returning home, Sam takes
> out the napkin and types the URI into a computer, which then retrieves
> the information to which Kim referred."
>
> If Kim use spaces within the URI then Sam cannot determine from the
> writing on the napkin how many spaces are used and at which places,
> making the transcription ambiguous.

That said, I agree that spaces should be avoided so that Kim can write research URL's on napkins.

Regards,
George
http://www.georgeevers.org

P.S. tomato, tomato, space, %20, URL, URI - more holistic is the napkin writ

Zooming back out to the broader concept, I still like the idea of recognizing basic HTML tags, which could be effectuated by an Insert Link button to assemble full <a href> tags for the user.

Regards,
George
http://www.georgeevers.org