Thread Subject: What's new in release 2010a?

I see that 2010a is available in pre-release, but I cannot find any information about what new features it introduces. Nor am I able to download the pre-release and try it for myself. Since it is now out in the open, can anyone say what's new and/or exciting about this release?

More than anything else I'm hoping for a new and revamped graphical system with better gui-building tools. I have heard some rumours for a long time that this was underway, but at 2009b I'm still struggling with its dreadful, primitive stone-age tools. So, any good news?

So is noone interested in the new release, or is it that one has to sign a non-disclosure agreement before downloading it? I mean, not a single soul is even willing to _speculate_ what's coming?

I would have thought the this to be a subject of some interest...

"Øyvind " <oyvist@gmail.com> wrote in message <hm8392$oa7$1@fred.mathworks.com>...
> So is noone interested in the new release, or is it that one has to sign a non-disclosure agreement before downloading it?

For me it's a later.

Bruno

"Øyvind " <oyvist@gmail.com> wrote in message <hm5f0s$53r$1@fred.mathworks.com>...
> I see that 2010a is available in pre-release, but I cannot find any information about what new features it introduces. Nor am I able to download the pre-release and try it for myself. Since it is now out in the open, can anyone say what's new and/or exciting about this release?
>
> More than anything else I'm hoping for a new and revamped graphical system with better gui-building tools. I have heard some rumours for a long time that this was underway, but at 2009b I'm still struggling with its dreadful, primitive stone-age tools. So, any good news?

I've been using it for all of 2 or 3 days now. No
problems so far, and it is reasonably fast.

bench
ans =
  0.047501 0.090149 0.14269 0.26638 0.38531 0.50896

I like to see the changes to fmincon, fixing some of
the problems that people have encountered over the
years. The new SQP algorithm:

  - Honors bounds at all iterations

  - Attempts a different step if one leads to an objective or constraint function returning a NaN, Inf, or complex result

  - Fast internal linear algebra for solving quadratic programs

The first two bullets are the ones that I care about
most.

The other thing I am happy to have is the 64 bit
version for the Mac OS. I've got one app right now
that I am playing with that will benefit greatly from
that ability.

Other than that, well its only been a couple of days.

John

"John D'Errico" <woodchips@rochester.rr.com> wrote in message <hm88i1$511$1@fred.mathworks.com>...
 
> I've been using it for all of 2 or 3 days now. No
> problems so far, and it is reasonably fast.
>
> bench
> ans =
> 0.047501 0.090149 0.14269 0.26638 0.38531 0.50896
>
> I like to see the changes to fmincon, fixing some of
> the problems that people have encountered over the
> years. The new SQP algorithm:
>
> - Honors bounds at all iterations
>
> - Attempts a different step if one leads to an objective or constraint function returning a NaN, Inf, or complex result
>
> - Fast internal linear algebra for solving quadratic programs
>
> The first two bullets are the ones that I care about
> most.
>
> The other thing I am happy to have is the 64 bit
> version for the Mac OS. I've got one app right now
> that I am playing with that will benefit greatly from
> that ability.
>
> Other than that, well its only been a couple of days.
>
> John

That's interesting. Is there any new syntax, or completely new features in the basic language itself? I mean, does the 'What's new..' section in the documentation have anything 'big'?

Oh, and is this version 7.10, or 8.0?

-- Øyvind

"Øyvind " <oyvist@gmail.com> wrote in message <hm8lnm$j2o$1@fred.mathworks.com>...

>
> That's interesting. Is there any new syntax, or completely new features in the basic language itself? I mean, does the 'What's new..' section in the documentation have anything 'big'?
>
> Oh, and is this version 7.10, or 8.0?
>
> -- Øyvind

It's v 7.10. I've been too busy to get into the details of the prerelease, and as already noted it's subject to an NDA, but it looks like nothing 'big' or groundbreaking but rather an incremental update with bug fixes, some new features and extensions to multithreaded functions, and deprecation of certain old functions and toolbox GUIs.

Øyvind wrote:
> So is noone interested in the new release, or is it that one has to sign
> a non-disclosure agreement before downloading it? I mean, not a single
> soul is even willing to _speculate_ what's coming?

The syntax of 'for' has been enhanced so that if it is followed by a
sequence of non-blank characters that includes the * or ? characters, or
by a quoted sequence of characters that includes the * or ? characters
(and may include blank), then the sequence of characters is treated as a
filename pattern that is to be globbed and the loop iterated over each
resulting file name. For example,

for fid = 'C:\Documents and Settings\Roberson\Matlab 2010a\datadir\*.wav'

would expand the pattern and assign each name in turn to fid.

The functionality of save has been enhanced such that if the first
string passed to save has the format of a valid variable name, and that
variable name exists in the routine that called save, then instead of
the string being used literally, the value of the variable will be
retrieved and used as the filename. This enhancement will support the
often-requested sequence, e.g.,

filename = 'Eig_results.mat';

save filename X


There will be a new optional parameter to cell(), zeros(), ones(),
char(), single(), double(), logical(), the int*() and uint*() and
typecast routines and various other routines such as diag(); the
optional parameter will be named 'index0', and it will cause the
resulting array to be indexed starting from 0 rather than from 1. In the
rare situation where the number of indices provided is exactly enough
that logical indexing might be inferred, logical indexing of such arrays
will only be used if the provided indices are of type logical (or a
subclass of logical); for arrays without the index0 attribute, logical
indexing will continue to be performed if all of the indices have
_value_ 0 or 1, no matter what type.


There will be a new set of parameters to system_dependent with names
beginning with 'dwim' (Do What I Mean). As per usual, the effects of
these parameters will be undocumented.

the save filename X and index0 thing is horrible. Breaks old code and the consistency of Matlab language. I hope it is and will never be true.


 > retrieved and used as the filename. This enhancement will support the
> often-requested sequence, e.g.,
>
> filename = 'Eig_results.mat';
>
> save filename X
>
> There will be a new optional parameter to cell(), zeros(), ones(),
> char(), single(), double(), logical(), the int*() and uint*() and
> typecast routines and various other routines such as diag(); the
> optional parameter will be named 'index0', and it will cause the
> resulting array to be indexed starting from 0 rather than from 1. In the
> rare situation where the number of indices provided is exactly enough
> that logical indexing might be inferred, logical indexing of such arrays
> will only be used if the provided indices are of type logical (or a
> subclass of logical); for arrays without the index0 attribute, logical
> indexing will continue to be performed if all of the indices have
> _value_ 0 or 1, no matter what type.
>

Siyi Deng wrote:
> the save filename X and index0 thing is horrible. Breaks old code and
> the consistency of Matlab language. I hope it is and will never be true.
>
>
> > retrieved and used as the filename. This enhancement will support the
>> often-requested sequence, e.g.,
>>
>> filename = 'Eig_results.mat';
>>
>> save filename X

Apparently, Mathworks analyzed a fair number of programs to determine what the
impact of the change to save() would be, and found that the only programs that
would break were some of the submissions to the programming contests. On the
other hand, Mathworks apparently found that 17.3% of the analyzed programs
started _working_ properly with the change (though most of the program authors
apparently hadn't noticed the fault in their code.) According to a leaked
tweet, Mathworks figures that this single change will reduce the volume of
cases to their support centre by 6.2% ... though a distressingly high number
of users are expected to still put through cases based upon not knowing that
they have to put quotes around filenames in assignments, sigh.

Dear Walter!

> >> filename = 'Eig_results.mat';
> >> save filename X
>
> Apparently, Mathworks analyzed a fair number of programs to determine what the
> impact of the change to save() would be, and found that the only programs that
> would break were some of the submissions to the programming contests.
> On the other hand, Mathworks apparently found that 17.3% of the analyzed programs
> started _working_ properly with the change (though most of the program authors
> apparently hadn't noticed the fault in their code.) According to a leaked
> tweet, Mathworks figures that this single change will reduce the volume of
> cases to their support centre by 6.2% ... though a distressingly high number
> of users are expected to still put through cases based upon not knowing that
> they have to put quotes around filenames in assignments, sigh.

This analysis might be strongly biased by the selection of the programs! If TMW looks in the pool of programs, which have been sent to the support, they are naturally faced mainly with buggy programs. If a programmer works seriously and reads the documentation of SAVE before using it, there will be no misusing and with a higher probability there will be no need to send the source to TMW!

If 17.3% and 6.2% are in any way meaningful numbers, it would be a good idea to modify the == operator, such that FIND(0:0.1:1 == 3) replies 4 as expected by nearly 90% of the Matlab newbies !?!

In my opinion changing SAVE in the mentioned way fixes 1000 trivial bugs, but causes 5 bugs, which are really hard to find even and especially for professional programmers who are working with Matlab for many years. It would be more consistent to replace the corresponding sentence in the documentation of SAVE:
  "save(savefile, 'v1') % Use when filename is stored in a variable"
==>
  "savefile = fullfile(tempdir, 'TestFile.mat');
  save(savefile, 'v1') % Use when filename is stored in the variable savefile
  Common pitfall: save filename X
  This saves X to the file called 'filename.mat' "

Index0 for zero based indices??
A really assume a joke. This would cause really serious changes in the Mex interface. How could a Mex script like a handmade FIND determine which index to return? Do I understand correctly, that this would invalidate a lot of Mex functions published in the FEX? I hope, that I miss the point, that this is just your speculation and your are in a funny mood today.

> for fid = 'C:\Documents and Settings\Roberson\Matlab 2010a\datadir\*.wav'
> would expand the pattern and assign each name in turn to fid.

Really??? Then this would change its behaviour dramatically and the result would depend on the current directory:
  for k = 'abcd*'; disp(k); end

I remember this thread:
  http://www.mathworks.com/matlabcentral/newsreader/view_thread/267352
which points to some serious bugs in fundamental functions. I deeply hope, that 2010A fixes more bugs than it adds. Adding new ambiguous functionalities like zero-based indices, magic SAVE and "intelligent" FOR would absolutely not help to increase the quality of Matlab and the programs written in this language.

Kind regards, Jan

Walter Roberson <roberson@hushmail.com> wrote in message <hm8ub6$m68$1@canopus.cc.umanitoba.ca>...
> Øyvind wrote:
> > So is noone interested in the new release, or is it that one has to sign
> > a non-disclosure agreement before downloading it? I mean, not a single
> > soul is even willing to _speculate_ what's coming?
>
> The syntax of 'for' has been enhanced so that if it is followed by a
> sequence of non-blank characters that includes the * or ? characters, or
> by a quoted sequence of characters that includes the * or ? characters
> (and may include blank), then the sequence of characters is treated as a
> filename pattern that is to be globbed and the loop iterated over each
> resulting file name. For example,
>
> for fid = 'C:\Documents and Settings\Roberson\Matlab 2010a\datadir\*.wav'
>
> would expand the pattern and assign each name in turn to fid.
>
> The functionality of save has been enhanced such that if the first
> string passed to save has the format of a valid variable name, and that
> variable name exists in the routine that called save, then instead of
> the string being used literally, the value of the variable will be
> retrieved and used as the filename. This enhancement will support the
> often-requested sequence, e.g.,
>
> filename = 'Eig_results.mat';
>
> save filename X
>
>
> There will be a new optional parameter to cell(), zeros(), ones(),
> char(), single(), double(), logical(), the int*() and uint*() and
> typecast routines and various other routines such as diag(); the
> optional parameter will be named 'index0', and it will cause the
> resulting array to be indexed starting from 0 rather than from 1. In the
> rare situation where the number of indices provided is exactly enough
> that logical indexing might be inferred, logical indexing of such arrays
> will only be used if the provided indices are of type logical (or a
> subclass of logical); for arrays without the index0 attribute, logical
> indexing will continue to be performed if all of the indices have
> _value_ 0 or 1, no matter what type.
>
>
> There will be a new set of parameters to system_dependent with names
> beginning with 'dwim' (Do What I Mean). As per usual, the effects of
> these parameters will be undocumented.

and - yes - wr forgot to mention the arrival of a new set of functions...

     help goto; % <- and look at its many siblings...

us

Two of my new favorites, which I am sure will help slow some of the traffic on cssm:


>>help domyhomework
DOMYHOMEWORK Automatic assignment completion.
     FNAME = DOMYHOMEWORK(DESCRIPTION,DESIREDGRADE) writes to an M-File, FNAME, the correct solution to the homework assignment given by the text of DESCRIPTION. The optional input, DESIREDGRADE, will take into account the desired level of MATLAB competence. These are to be capital letters chose from 'A','B','C',D','F'. If 'A' is chose, the code within the M-File will be fully documented, with help and an H1 line, with input and error checking, and vectorized where appropriate for optimum efficiency. If 'D' is chosen, there will be no comments, no help, and nearly every line will use a call to EVAL. Also the code may be buggy.

See also DOMYSENIORPROJECT, DOMYGRADUATETHESIS

Dear Matt!

> Two of my new favorites, which I am sure will help slow some of the traffic on cssm:
> >>help domyhomework

And "goto" ?! You guys are making jokes and I took Walter almost for serious?!
I should stop posting on weekends and reading posts about future Matlab releases, before I get nightmares.

Jan

"Jan Simon"
> Dear Matt!
>
> > Two of my new favorites, which I am sure will help slow some of the traffic on cssm:
> > >>help domyhomework
>
> And "goto" ?! You guys are making jokes and I took Walter almost for serious?!
> I should stop posting on weekends and reading posts about future Matlab releases, before I get nightmares.
>
> Jan

Almost?!! Do you define a full-page answer, links and examples included, as ALMOST?????

 BTW, I'm a bit disappointed that in the SEE ALSO line of DOMYHOMEWORK there's no mention of the most awesome function: ifYouSeekAyou.

Oleg

Dear Oleg!
 
> > And "goto" ?! You guys are making jokes and I took Walter almost for serious?!
>
> Almost?!! Do you define a full-page answer, links and examples included, as ALMOST?????

While GOTO is obviously funny, I believe that TMW is able to implement Index0 and a do-what-you-think-that-I-mean-SAVE.
Anyhow, system_dependent('dwim') ?! This is working since 2008a, so what?

Silly me, Jan

Øyvind wrote:
> So is noone interested in the new release, or is it that one has to sign
> a non-disclosure agreement before downloading it? I mean, not a single
> soul is even willing to _speculate_ what's coming?

A new functionality is to be introduced for sort(). sort() will continue
to support the current forms but in addition it will support adding a
function handle after the usual parameters (though before or after the
optional direction keyword is supported). For example,

sort(A,2,@mycompare,'descend')

The function handle will be expected to take two parameters of the same
type as class(A), and to return -1, 0, or 1, indicating whether the
first parameter is less than, equal to, or greater than the second
parameter.

In this way, sort is extended to support custom sort orders, and to
support sorting of non-numeric (or char) objects such as objects or cell
arrays.

Walter Roberson <roberson@hushmail.com> wrote in message <hmc9n5$c3b$1@canopus.cc.umanitoba.ca>...

>
> A new functionality is to be introduced for sort(). sort() will continue
> to support the current forms but in addition it will support adding a
> function handle after the usual parameters (though before or after the
> optional direction keyword is supported). For example,
>
> sort(A,2,@mycompare,'descend')
>

Seriously, it is not had to write one. The issue is while SORT on numeric array uses processor arithmetic comparison that is far faster than calling any Matlab comparison function call which is - I expect - MUCH slower due to overhead. That alone defeats any attempt to implement any efficient sorting with customized order. The best alternative is up for user to map his objects to numerical array before sorting.

Bruno

"Matt Fig" <spamanon@yahoo.com> wrote in message <hmb8dc$8fi$1@fred.mathworks.com>...
> Two of my new favorites, which I am sure will help slow some of the traffic on cssm:
>
>
> >>help domyhomework
> DOMYHOMEWORK Automatic assignment completion.
> FNAME = DOMYHOMEWORK(DESCRIPTION,DESIREDGRADE) writes to an M-File, FNAME, the correct solution to the homework assignment given by the text of DESCRIPTION. The optional input, DESIREDGRADE, will take into account the desired level of MATLAB competence. These are to be capital letters chose from 'A','B','C',D','F'. If 'A' is chose, the code within the M-File will be fully documented, with help and an H1 line, with input and error checking, and vectorized where appropriate for optimum efficiency. If 'D' is chosen, there will be no comments, no help, and nearly every line will use a call to EVAL. Also the code may be buggy.
>
> See also DOMYSENIORPROJECT, DOMYGRADUATETHESIS

Shouldn't the parameters of the DOMYHOMEWORK function be reversed? I thought it should be

FNAME = DOMYHOMEWORK(DESIREDGRADE,DESCRIPTION)

where DESCRIPTION is an optional parameter to allow it not to be used so that students need not break into valuable drinking time actually describing what he/she wants the application to do - after all thats what a significant number of these students expect us to work with. This should be relatively easy for TMW to implement using the Steve Eddins Mind Reading function - surely he must have it perfected by now?

Regards

Dave Robinson

Dave Robinson wrote:
> "Matt Fig" <spamanon@yahoo.com> wrote in message
> <hmb8dc$8fi$1@fred.mathworks.com>...
>> Two of my new favorites, which I am sure will help slow some of the
>> traffic on cssm:
>>
>>
>> >>help domyhomework
>> DOMYHOMEWORK Automatic assignment completion.
>> FNAME = DOMYHOMEWORK(DESCRIPTION,DESIREDGRADE) writes to an
>> M-File, FNAME, the correct solution to the homework assignment given
>> by the text of DESCRIPTION. The optional input, DESIREDGRADE, will
>> take into account the desired level of MATLAB competence. These are
>> to be capital letters chose from 'A','B','C',D','F'. If 'A' is chose,
>> the code within the M-File will be fully documented, with help and an
>> H1 line, with input and error checking, and vectorized where
>> appropriate for optimum efficiency. If 'D' is chosen, there will be
>> no comments, no help, and nearly every line will use a call to EVAL.
>> Also the code may be buggy.
>> See also DOMYSENIORPROJECT, DOMYGRADUATETHESIS
>
> Shouldn't the parameters of the DOMYHOMEWORK function be reversed? I
> thought it should be
> FNAME = DOMYHOMEWORK(DESIREDGRADE,DESCRIPTION)
>
> where DESCRIPTION is an optional parameter to allow it not to be used so
> that students need not break into valuable drinking time actually
> describing what he/she wants the application to do - after all thats
> what a significant number of these students expect us to work with. This
> should be relatively easy for TMW to implement using the Steve Eddins
> Mind Reading function - surely he must have it perfected by now?
>
> Regards
>
> Dave Robinson

Boy, you guys have long memories. ;-)

Still working on it, I'm afraid.

---
Steve Eddins
http://blogs.mathworks.com/steve/

Steve Eddins <Steve.Eddins@mathworks.com> wrote in message
>
> Boy, you guys have long memories. ;-)
>
> Still working on it, I'm afraid.
>
> ---
> Steve Eddins
> http://blogs.mathworks.com/steve/



Well, it surely would be among the most important M-Files ever written!

"Matt Fig"
> Well, it surely would be among the most important M-Files ever written!

but then - the HE man promised this almost exactly ten(10) years ago - and still running(?)...
STEVE(!)... where -exactly- are we...

http://groups.google.com/group/comp.soft-sys.matlab/msg/37917d2b0ee5e7b1?dmode=source

us

"Dave Robinson" <dave.robinson@somewhere.biz> wrote in message
news:hmdmgn$c0v$1@fred.mathworks.com...
> "Matt Fig" <spamanon@yahoo.com> wrote in message
> <hmb8dc$8fi$1@fred.mathworks.com>...
>> Two of my new favorites, which I am sure will help slow some of the
>> traffic on cssm:
>>
>>
>> >>help domyhomework
>> DOMYHOMEWORK Automatic assignment completion.
>> FNAME = DOMYHOMEWORK(DESCRIPTION,DESIREDGRADE) writes to an M-File,
>> FNAME, the correct solution to the homework assignment given by the text
>> of DESCRIPTION. The optional input, DESIREDGRADE, will take into account
>> the desired level of MATLAB competence. These are to be capital letters
>> chose from 'A','B','C',D','F'. If 'A' is chose, the code within the
>> M-File will be fully documented, with help and an H1 line, with input and
>> error checking, and vectorized where appropriate for optimum efficiency.
>> If 'D' is chosen, there will be no comments, no help, and nearly every
>> line will use a call to EVAL. Also the code may be buggy. See also
>> DOMYSENIORPROJECT, DOMYGRADUATETHESIS
>
> Shouldn't the parameters of the DOMYHOMEWORK function be reversed? I
> thought it should be
> FNAME = DOMYHOMEWORK(DESIREDGRADE,DESCRIPTION)
>
> where DESCRIPTION is an optional parameter to allow it not to be used so
> that students need not break into valuable drinking time actually
> describing what he/she wants the application to do - after all thats what
> a significant number of these students expect us to work with. This should
> be relatively easy for TMW to implement using the Steve Eddins Mind
> Reading function - surely he must have it perfected by now?

Well, I shouldn't be posting this, but here's the example from the M-file
help of DOMYHOMEWORK:

    functionName = doMyHomework('A',
urlread('http://www.someuniversity.edu/ma123/assignment1.html'))

So it shouldn't cut down on students' valuable drinking time _too_ much.

Hmm ... now that I think of it, I wonder if we were able to fix that "bug"
where the desiredgrade parameter was always set to 'F' inside the function
itself, ignoring the user's input? Oh well, if we didn't I'm sure we'll
hear about it sooner or later :)

--
Steve Lord
slord@mathworks.com
comp.soft-sys.matlab (CSSM) FAQ: http://matlabwiki.mathworks.com/MATLAB_FAQ

"Steven Lord" <slord@mathworks.com> wrote in message
> Well, I shouldn't be posting this, but here's the example from the M-file
> help of DOMYHOMEWORK:
>
> functionName = doMyHomework('A',
> urlread('http://www.someuniversity.edu/ma123/assignment1.html'))
>
> So it shouldn't cut down on students' valuable drinking time _too_ much.
>
> Hmm ... now that I think of it, I wonder if we were able to fix that "bug"
> where the desiredgrade parameter was always set to 'F' inside the function
> itself, ignoring the user's input? Oh well, if we didn't I'm sure we'll
> hear about it sooner or later :)
>
> --
> Steve Lord
> slord@mathworks.com
> comp.soft-sys.matlab (CSSM) FAQ: http://matlabwiki.mathworks.com/MATLAB_FAQ
>

All this is very nice... But still no new gui-building tools?

Seriously, though, is there some custom that one never seriously discusses future matlab-versions on this forum? While some freewheeling humorous speculation (which I sort of asked for in my first post) is fun, is the future direction of Matlab of little interest to its users?

Dear Øyvind!

> Seriously, though, is there some custom that one never seriously discusses future matlab-versions on this forum? While some freewheeling humorous speculation (which I sort of asked for in my first post) is fun, is the future direction of Matlab of little interest to its users?

I *was* serious. I am alone, I am not tall, I am not a native speaker, but I am here.

For my business stability and reliability is much more important than new features. Even the old-fashioned GUI tools would not bother me, if they at least would not crash. As long as TMW cannot avoid bugs like e.g. the "A'\b" problem in 2009b, future features are more or less meaningless for me.

I send bug reports and ideas for improvements to TMW. I publish these ideas in the FEX, if the developpers do not react for 6 month (and if this is no conflict with the license conditions). My bug report for ANCESTOR has been rejected at first, because I did not attach my license number - how funny!

Kind regards, Jan

Øyvind wrote:

> Seriously, though, is there some custom that one never seriously
> discusses future matlab-versions on this forum? While some freewheeling
> humorous speculation (which I sort of asked for in my first post) is
> fun, is the future direction of Matlab of little interest to its users?

I haven't ever checked Mathwork's beta-test and pre-release licenses,
but on every company I've tested for, beta and pre-release testers are
under non-disclosure agreements. If I knew what was coming in 2010a (or
any future release), I would not be allowed to tell you.

I do test software for companies from time to time, but I only publicly
talk about officially released versions. If I were the kind of person
who broke non-disclosure agreements, I would not deserve my security
clearance.

Walter Roberson <roberson@hushmail.com> wrote in message <hmjcto$bdv$1@canopus.cc.umanitoba.ca>...
> Øyvind wrote:
>
> > Seriously, though, is there some custom that one never seriously
> > discusses future matlab-versions on this forum? While some freewheeling
> > humorous speculation (which I sort of asked for in my first post) is
> > fun, is the future direction of Matlab of little interest to its users?
>
> I haven't ever checked Mathwork's beta-test and pre-release licenses,
> but on every company I've tested for, beta and pre-release testers are
> under non-disclosure agreements. If I knew what was coming in 2010a (or
> any future release), I would not be allowed to tell you.
>
> I do test software for companies from time to time, but I only publicly
> talk about officially released versions. If I were the kind of person
> who broke non-disclosure agreements, I would not deserve my security
> clearance.

I think this is a perfectly fine answer. In fact I asked about NDAs earlier, but the replies were mixed, and though some people were obviously joking, others seemed to be answering in earnest. I don't think you can assume that the practice with NDAs is blindingly obvious to everyone.

So no information on r2010a is publicly available as of yet. Okidokey.

I do find that in general there seems to be little discussion of new features in this forum, even after public release. While many, quite understandably, care more about stability, bugs, etc. it's a bit surprising (statistically) that with such a vast userbase there isn't much 'buzz' surrounding 'new stuff'. (As I don't contribute much myself, I'm in no position to complain, it's more of an observation.)

In article <hmlapl$6e4$1@fred.mathworks.com>,
? yvind <oyvist@gmail.com> wrote:
>I do find that in general there seems to be little discussion of new
>features in this forum, even after public release. While many, quite
>understandably, care more about stability, bugs, etc. it's a bit
>surprising (statistically) that with such a vast userbase there isn't
>much 'buzz' surrounding 'new stuff'. (As I don't contribute much myself,
>I'm in no position to complain, it's more of an observation.)

Agreed.

Francis

R2010a is now available. You can see what's new from
http://www.mathworks.com/

"Pekka Kumpulainen" <pekka.nospam.kumpulainen@tut.please.fi> wrote in message <hmqkhk$92j$1@fred.mathworks.com>...
> R2010a is now available. You can see what's new from
> http://www.mathworks.com/

Thanks. Quickly scanned the release notes, and there seem to be mostly minor changes.

Absolutely dreadfully disappointed (as usual) that there is no update to the GUI system. I need to start looking around for something other than Matlab, but the task of learning a new language and porting my project is extremely daunting...

Aargh.

Øyvind wrote:
> Absolutely dreadfully disappointed (as usual) that there is no update to
> the GUI system.

What kinds of changes to the GUI system were you hoping for? What do
you find difficult or impossible now?


--
Doug Schwarz
dmschwarz&ieee,org
Make obvious changes to get real email address.

Jumping in with a few quick thoughts:

Øyvind - sorry to hear about your disappointment with the lack of enhancements to the GUI building system. We know that you aren't the only user who is waiting anxiously for an upgrade to our capabilities in this area.

The feature that I'm REALLY excited about is buried in some pretty obscure text on the release highlight page:
"New System objects for stream processing in MATLAB, with over 140 supported algorithms in Video and Image Processing Blockset and Signal Processing Blockset"

Short story: You can now use some Simulink blocks directly in MATLAB (Simulink not required).

Longer story: If you ever program time- or sample-based algorithms in MATLAB, this introduces a way-better programming model. Ditto for more generic stream processing, like looping through data in an arbitrarily large file. States and incrementing are managed for you, allowing you to write code that looks something like this, where I manage a filter, audio input, and audio output as 3 of these objects Filter, AudioIn, and AudioOut.

while ~isDone(AudioIn)
  sig = step(AudioIn);
  filteroutput = step(Filter, sig);
  step(AudioOut, filteroutput);
end

It's not my bag, but these also play with Fixed Point, and to some extent with embedded MATLAB, allowing to convert your MATLAB code using system objects to C code.

- scott

"Øyvind " <oyvist@gmail.com> wrote in message <hmqm1a$cnv$1@fred.mathworks.com>...
> "Pekka Kumpulainen" <pekka.nospam.kumpulainen@tut.please.fi> wrote in message <hmqkhk$92j$1@fred.mathworks.com>...
> > R2010a is now available. You can see what's new from
> > http://www.mathworks.com/
>
> Thanks. Quickly scanned the release notes, and there seem to be mostly minor changes.
>
> Absolutely dreadfully disappointed (as usual) that there is no update to the GUI system. I need to start looking around for something other than Matlab, but the task of learning a new language and porting my project is extremely daunting...
>
> Aargh.

Hi TMW,

While on the topic of Matlab GUI's, I would like to suggest adding if not available (and, if available, to make them easier to implement) ...

1) the functionality of having multiple lines of text for static text objects.
--Rationale: I have often wanted to have two or three lines of descriptive static text in a static text box, however, I have not found it possible to do so. For example, maybe I have only a constrained amount of horizontal space to work with, but I have plenty of vertical space to wrap text into. I have looked into using the max/min properties for the static text box in guide, but it appears that this property does not work for static text boxes, only for editable text boxes.

2) the functionality of being able to rotate gui objects in guide.
--Rationale: I have often wanted to rotate a piece of static text to be vertical in guide, rather than horizontal. This should be a feature that is available.

3) colorbar outside a given guide axis.
--Rationale: I have often had an image in a given guide axis that perhaps I want a colorbar next to. The problem is that the colorbar appears inside the axis and shrinks my image, messes up the alignment of scrollbars wrt the actual image, and then when I get rid of the colorbar my image is still shrunk. I have tried to set the colorbar properties to have location eastOutside, but it doesn't help. I would like to see the colorbar outside the guide axis, in the empty space outside the axis. Otherwise, when designing your gui, how might one plan for this image shrinking and location of colorbars, easily. Perhaps, this isn't the right place for this comment, but I think that this is another example of something that TMW should expect the user might want.

4) mixing of logical interactivity with numeric matrix in uitable
--Rationale: I have found that when I have numeric data coming into two columns that is non-editable and when I want a 3rd column that is logical and editable, then matlab does not allow me to edit the logical checkbox and returns an error "cannot convert logical edit to numeric matrix". This is obvious but also a nuisance.

OK, well, I think 4 examples is sufficient for now.

“Imagination is more important than knowledge. For knowledge is limited to all we now know and understand, while imagination embraces the entire world, and all there ever will be to know and understand.” -- albert einstein

Eric Diaz

In article <hn132t$97s$1@fred.mathworks.com>,
 "Eric Diaz" <eric.diaz@gmail.com> wrote:

[snip]

> 1) the functionality of having multiple lines of text for static text
> objects.

Use a cell array of strings for the String property.

[snip]


> 2) the functionality of being able to rotate gui objects in guide.
> --Rationale: I have often wanted to rotate a piece of static text to be
> vertical in guide, rather than horizontal. This should be a feature that is
> available.

This is a good suggestion and I have added it to my uibutton function.
(In spite of the name, it can be used to create something that looks
like a text uicontrol.) I just uploaded the new version.

<http://www.mathworks.com/matlabcentral/fileexchange/10743>

--
Doug Schwarz
dmschwarz&ieee,org
Make obvious changes to get real email address.

Hye Doug,

Thanks for your input. I thought I had tried that "cell array" technique in the typical staticTextBox_callback or create_fxn, but to no avail though. I guess I can try it again.

Two thumbs up on your uibutton function though!,

Cheers,

Eric Diaz

Doug Schwarz <see@sig.for.address.edu> wrote in message <see-7311E7.18493607032010@news.frontiernet.net>...
> In article <hn132t$97s$1@fred.mathworks.com>,
> "Eric Diaz" <eric.diaz@gmail.com> wrote:
>
> [snip]
>
> > 1) the functionality of having multiple lines of text for static text
> > objects.
>
> Use a cell array of strings for the String property.
>
> [snip]
>
>
> > 2) the functionality of being able to rotate gui objects in guide.
> > --Rationale: I have often wanted to rotate a piece of static text to be
> > vertical in guide, rather than horizontal. This should be a feature that is
> > available.
>
> This is a good suggestion and I have added it to my uibutton function.
> (In spite of the name, it can be used to create something that looks
> like a text uicontrol.) I just uploaded the new version.
>
> <http://www.mathworks.com/matlabcentral/fileexchange/10743>
>
> --
> Doug Schwarz
> dmschwarz&ieee,org
> Make obvious changes to get real email address.

Doug Schwarz <see@sig.for.address.edu> wrote in message <Qfbkn.56021$0N3.23829@newsfe09.iad>...
> Øyvind wrote:
> > Absolutely dreadfully disappointed (as usual) that there is no update to
> > the GUI system.
>
> What kinds of changes to the GUI system were you hoping for? What do
> you find difficult or impossible now?
>
>
> --
> Doug Schwarz
> dmschwarz&ieee,org
> Make obvious changes to get real email address.


OK, this is slightly rambling, and just touches on my two main objections to the GUI system in Matlab.

I have little programming experience beyound Matlab, but my general impression when I look around is that other programming environments have simply much more advanced capabilities than Matlab, without compromising ease of use. (My one time trying NetBeans I made a beautiful GUI in 10 minutes starting from scratch with no previous experience, and not understanding java. GUIs made by GUIDE always come out ugly and awkward-looking. Also, it is difficult to get a uniform 'look' across GUIs that belong together--border sizes, button sizes, etc.)

A short wish list:

1)
A better gui designer than GUIDE, which has no real layout management. You get to place uicontrols absolutely, specified in pixels or characters, or 'normalized'. The resizing behaviour is atrocious. Writing a ResizeFcn is painstaking, and difficult to adapt to changes in the GUI.

The production of a .fig-file is unappealing, I'd like the layout to be in pure matlab-code. The alternative is to write the layout code by hand, which is difficult and tedious, and making small changes later (like adding an uicontrol) can be time-consuming.

There exist some user-contributed layout managers, like GridBagLayout, but these are also difficult to use, feel a bit 'fragile', and slows down execution speed.

So I'd like a good gui designer which produces Matlab code and uses proper layout management.


2)
Improved uicontrols. Right now they seem to be gutted swing components, with most of the functionality removed. This is done to make them easier to use, I'm sure, but in some cases it removes essential functionality.

Common widgets, like spinners, are unavailable. You can add them in using java code, but the interface is hard to figure out if you have no java experience, and there is little documentation.

If you are tempted to use undocumented features of Matlab to enhance your GUIs, Mathworks keeps threatening to remove or change functionality without warning.

I prefer a more object-oriented approach to the current set/get-interface. You can get this by using java or the 'handle' function, but this is awkward, and the 'handle' function is undocumented and can be changed in future versions.

I was waiting for the batch license file generation for TAH license finally. Doesn't seem to be possible yet.
Unfortunately the VPN bug in windows hasn't been fixed either. Or fortunately. Now we don't need to make those 150 license files for this version either. Back to 2009a.

In article <hn2c1j$m8i$1@fred.mathworks.com>,
? yvind <oyvist@gmail.com> wrote:
>OK, this is slightly rambling, and just touches on my two main
>objections to the GUI system in Matlab.
>
>I have little programming experience beyound Matlab, but my general
>impression when I look around is that other programming environments
>have simply much more advanced capabilities than Matlab, without
>compromising ease of use. (My one time trying NetBeans I made a
>beautiful GUI in 10 minutes starting from scratch with no previous
>experience, and not understanding java. GUIs made by GUIDE always come
>out ugly and awkward-looking. Also, it is difficult to get a uniform
>'look' across GUIs that belong together--border sizes, button sizes,
>etc.)

I program in Delphi (Object Pascal) and the 'GUI builder' that
is part of the IDE has a very solid feel with very few surprises.
So, for example, if I select a control and fine adjust its position
by pressing keys, each step is the same (a screen pixel). In GUIDE,
the movements appear inconsistent and 'jumpy'. I have no idea why
this is, but it *feels* fragile. The property editor feels more
responsive somehow too; in contrast, MATLAB's feels 'clunky'.

>[...]
>I prefer a more object-oriented approach to the current
>set/get-interface. You can get this by using java or the 'handle'
>function, but this is awkward, and the 'handle' function is undocumented
>and can be changed in future versions.

Having properties get and set with a simple syntax like

RadioButton1.String = 'Push me'

would be wonderful.

I still get grief from passing data around inside and between
GUIs, even there are so many alternative ways of doing it. My
main gripe is having to ensure that 'handles' is included as a
function argument whenever UI controls need to be accessed. It
would be nice if they had file (or user-settable) scope.

TBH, my ideal programming environment would be the Delphi IDE
with VCL-like controls, but with MATLAB's array syntax and
access to all the wonderful stuff that MATLAB can do in one
or two lines.

Francis

fburton@nyx.net (Francis Burton) wrote in message <1268049371.9660@irys.nyx.net>...
> Having properties get and set with a simple syntax like
>
> RadioButton1.String = 'Push me'
>
> would be wonderful.

You can actually get this by using the undocumented function 'handle' on the graphics handle, like this:
>> RadioButton1 = handle(RadioButton1);
Then you have dot-syntax for setting and getting properties, and also tab-completion.

Unfortunately it doesn't work recursively down through a hierarchy of handles, so that e.g. axes.Children returns numerical handles, even though 'axes' has been operated on by 'handle'. Also, undocumeted features like this may suddenly stop working in future versions--no warning.

> I still get grief from passing data around inside and between
> GUIs, even there are so many alternative ways of doing it. My
> main gripe is having to ensure that 'handles' is included as a
> function argument whenever UI controls need to be accessed. It
> would be nice if they had file (or user-settable) scope.

The way I make GUIs, I use GUIDE for the layout (although I've tried coding by hand quite a few times), then convert the fig-file to matlab-code using e.g. http://www.mathworks.ch/matlabcentral/fileexchange/14340-convert-fig-to-m
Then implement the callbacks using nested functions. It is also possible to just load the fig-file inside the m-file and then implement the callbacks. On no account would I let GUIDE generate the callbacks for me! You just need to look at the mlint report for GUIDE-generated files--they are absolutely littered with warnings.

"Øyvind " <oyvist@gmail.com> wrote in message <hmqm1a$cnv$1@fred.mathworks.com>...

> Thanks. Quickly scanned the release notes, and there seem to be mostly minor changes.
>
> Absolutely dreadfully disappointed (as usual) that there is no update to the GUI system. I need to start looking around for something other than Matlab, but the task of learning a new language and porting my project is extremely daunting...
>
> Aargh.

Couldn't agree more, but our focus is on the general graphics rather than the GUI:

- No native full-color (i.e. one colormap per figure), although there are quite useful FEX contributions.

- Ticklabels without much flexibility at all, and a plethora of FEX submissions to help out.

- Colorbar that is rather difficult to customise (same for legend, I am told), too much going on under the hood because it tries but dos not manage to be intelligent.

- the new BOXPLOT ticklabels are a veritable nightmare to customise, because they are not actually ticklabels but some text object hidden deep (deep!!) down in the BOXPLOT handle graphics tree, but with strings (listeners) attached that tries to outsmart you by restoring whatever changes. Maybe useful behaviour if one only wants a quicks, straight and simple boxplot, but for anything else, well I say no more ....

The BOXPLOT (mis-)behaviour is fairly new, but for the other things I have for a long time been putting in suggestions/requests through various channels, but to absolutely no effect.

Thus, since about a year we have begun moving most our serious and more demanding plotting tasks to other software.

Lars

Hi Doug,

Still can't get multiple lines of text on a static guide text object. I tried using a cell array of strings (actually using the java example in cell help & using the more difficult cellstr function) in the static text create function but it doesn't work.


% --- Executes during object creation, after setting all properties.
function StaticImagePushButton_CreateFcn(hObject, eventdata, handles)
% hObject handle to StaticImagePushButton (see GCBO)
% eventdata reserved - to be defined in a future version of MATLAB
% handles empty - handles not created until after all CreateFcns called

Method #1
strArray = java_array('java.lang.String', 2);
strArray(1) = java.lang.String('Load');
strArray(2) = java.lang.String('Static Image');
cellArray = cell(strArray);
set(hObject,'min',0,'max',2,'string',cellArray);

Method #2
strArray = ['Load ';'Static Images'];
cellArray = cellstr(strArray);
set(hObject,'min',0,'max',2,'string',cellArray);

I looked at your uibutton gui but it doesn't really seem intuitive with respect to its usage within guide.

I think that TMW should really fix this functionality.

Eric Diaz

> Still can't get multiple lines of text on a static guide text object. I tried using a cell array of strings (actually using the java example in cell help & using the more difficult cellstr function) in the static text create function but it doesn't work.


Multi-line text labels are unsupported in the standard uicontrols, but you can use simple Java controls and HTML line-breaks for this:

jLabel = javax.swing.JLabel('<html>This is a long<br>multi-line string');
[jhLabel,hcontainer] = javacomponent(jLabel,[10,10,100,30],gcf);

If your multiple lines are stored in a cell-array, you can convert this into a multi-line HTML string as follows:

strs = {'very long','multi-line','string'};
htmlStr = sprintf('<html>%s%s', sprintf('%s<br>',c{1:end-1}), c{end});
jLabel = javax.swing.JLabel(htmlStr);
[hcomponent,hcontainer] = javacomponent(jLabel,[10,50,50,20],gcf);

More about using HTML in Matlab labels: http://undocumentedmatlab.com/blog/customizing-matlab-labels/

Yair Altman
http://UndocumentedMatlab.com

Eric Diaz wrote:
> Hi Doug,
>
> Still can't get multiple lines of text on a static guide text object. I
> tried using a cell array of strings (actually using the java example in
> cell help & using the more difficult cellstr function) in the static
> text create function but it doesn't work.

Eric,

You must be using an older version -- it works in R2009a at least. It's
even documented.


> I looked at your uibutton gui but it doesn't really seem intuitive with
> respect to its usage within guide.

I guess I'm not too surprised uibutton doesn't play well with guide as I
have never used it. It was worth a shot.

--
Doug Schwarz
dmschwarz&ieee,org
Make obvious changes to get real email address.

Hi Doug,

Actually, I have MATLAB Version 7.9.0.529 (R2009b). So, maybe the functionality broke between 2009a and 2009b.

Thanks,

Eric

Doug Schwarz <see@sig.for.address.edu> wrote in message <j2Tnn.22381$NH1.5698@newsfe14.iad>...
> Eric Diaz wrote:
> > Hi Doug,
> >
> > Still can't get multiple lines of text on a static guide text object. I
> > tried using a cell array of strings (actually using the java example in
> > cell help & using the more difficult cellstr function) in the static
> > text create function but it doesn't work.
>
> Eric,
>
> You must be using an older version -- it works in R2009a at least. It's
> even documented.
>
>
> > I looked at your uibutton gui but it doesn't really seem intuitive with
> > respect to its usage within guide.
>
> I guess I'm not too surprised uibutton doesn't play well with guide as I
> have never used it. It was worth a shot.
>
> --
> Doug Schwarz
> dmschwarz&ieee,org
> Make obvious changes to get real email address.

Hi Yair,

Thanks for your input. I had actually seen the html stuff for guide on the website before when I had been looking, however, since I don't know html or java hardly at all, I was just lost.

I appreciate that it there is a lot of documented undocumented functionality though on the website!

Thanks again,

Eric

"Yair Altman" <altmanyDEL@gmailDEL.comDEL> wrote in message <hnot9n$34o$1@fred.mathworks.com>...
> > Still can't get multiple lines of text on a static guide text object. I tried using a cell array of strings (actually using the java example in cell help & using the more difficult cellstr function) in the static text create function but it doesn't work.
>
>
> Multi-line text labels are unsupported in the standard uicontrols, but you can use simple Java controls and HTML line-breaks for this:
>
> jLabel = javax.swing.JLabel('<html>This is a long<br>multi-line string');
> [jhLabel,hcontainer] = javacomponent(jLabel,[10,10,100,30],gcf);
>
> If your multiple lines are stored in a cell-array, you can convert this into a multi-line HTML string as follows:
>
> strs = {'very long','multi-line','string'};
> htmlStr = sprintf('<html>%s%s', sprintf('%s<br>',c{1:end-1}), c{end});
> jLabel = javax.swing.JLabel(htmlStr);
> [hcomponent,hcontainer] = javacomponent(jLabel,[10,50,50,20],gcf);
>
> More about using HTML in Matlab labels: http://undocumentedmatlab.com/blog/customizing-matlab-labels/
>
> Yair Altman
> http://UndocumentedMatlab.com

"Eric Diaz" <eric.diaz@gmail.com> wrote in message <hnor5d$smu$1@fred.mathworks.com>...
> Hi Doug,
>
> Still can't get multiple lines of text on a static guide text object. I tried using a cell array of strings (actually using the java example in cell help & using the more difficult cellstr function) in the static text create function but it doesn't work.
>
>
> % --- Executes during object creation, after setting all properties.
> function StaticImagePushButton_CreateFcn(hObject, eventdata, handles)
> % hObject handle to StaticImagePushButton (see GCBO)
> % eventdata reserved - to be defined in a future version of MATLAB
> % handles empty - handles not created until after all CreateFcns called
>
> Method #1
> strArray = java_array('java.lang.String', 2);
> strArray(1) = java.lang.String('Load');
> strArray(2) = java.lang.String('Static Image');
> cellArray = cell(strArray);
> set(hObject,'min',0,'max',2,'string',cellArray);
>
> Method #2
> strArray = ['Load ';'Static Images'];
> cellArray = cellstr(strArray);
> set(hObject,'min',0,'max',2,'string',cellArray);
>
> I looked at your uibutton gui but it doesn't really seem intuitive with respect to its usage within guide.
>
> I think that TMW should really fix this functionality.

........................................

Currently working on project and m using the program "digit recognition" as part of it. I'm stuck with the last m-file 'm suppose to use..digitregui.m where:
- % eventdata reserved - to be defined in a future version of MATLAB.

Is it defined in Matlab 2010a?? Currently using 2007 version!