help-octave
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: using binary octave 2.9.12


From: David Bateman
Subject: Re: using binary octave 2.9.12
Date: Mon, 18 Jun 2007 10:20:40 +0200
User-agent: Thunderbird 1.5.0.7 (X11/20060921)

Robert A. Macy wrote:
> I'm kind of stuck with Windows for several projects so
> really appreciated when I found some kind soul had done a
> new binary for the recent version of octave
>
> Is there a place, or directly to someone, I can make
> comments?  I'm sure most of what I see is repairable by a
> knowledgeable SW guy.  
>
> For examples, 
> 1) octave 2.1.50a was 7+MB, octave 2.9.12 is 30+MB, which
> is ok, I did find more in it.  But I'm really learning to
> hate the tendency towards bloated code.
>   
Do you want sparse matrices, non-linear solvers, zipped load/save code,
regexp? All of these add both code to Octave and external dependencies
over what was in 2.1.50.. What about NDArrays, integer types? These also
add significant code.. In any case if you consider these additions
bloat, there are many others who'd disagree with you. The only thing I
consider bloat is when a piece of code is written badly and might be
written more compactly. Even in the case of real bloat (I can't think of
any of the top of my head), there might be reasons to keep it, like it
is code that is known to work and changing it would result in
significant instability during a period.

> 2) Takes a long time to start, well ok there's more code.
>   
Type "pkg list" and see the packages that are automatically loaded.. I
expect 2.9.13 to load quite a bit faster, and you also might improve the
speed by doing "pkg rebuild -noauto <package>" on the packages that you
don't use.

> 3) The little window that opens is...well, white letters on
> a black background, looks like command-prompt, very
> difficult to read.  worse, the window cannot be expanded in
> width and most of what is left available is taken up with a
> very long folder description and '>'.  When using octave, I
> write very long concatenated lines to work real-time.
>  Luckily the screen resolution allows both plot and window
> without overlap.  Also, any plot function removes the
> script window from being active. so instead of type
> something new and keep watching the plot change, I have to
> plot, move my hand to mouse and click on script window,
> type and look at plot, move hand to mouse and click on
> script window, type etc etc  the old 2.1.50a I could sit
> and just type new constructions over and over and never
> have to reach for that stupid mouse.  
>   
It is a command window.. The command window is an integral part of what
Octave is.. This might be hidden in an IDE as things like OctaveWorkshop
or QtOctave have tried to do. But even then it is just a command window
with a few bells and whistles.

I'm not a windows users, but it would surprise me if cmd.exe was not
configurable. As for the prompt, look at the "PS1" variable and you
might want to add a PS1 seeting to your octaverc file. This is a gnuplot
thing.. It also can be configured to prevent it stealing the focus.
Search this list for the last two weeks.

Please note 2.9.12 release is an experimental release. Ask Michael
politely and he might include the variable to force gnuplot not to steal
the focus.

> 4) Octave appeared to run significantly slower than my old
> 2.1.50a.  I was going to do some benchmark tests using
> scripts and what I normally do.  but I ran into further
> snags with the plot program.  Takes 5 seconds to come on.
> and if 3d, takes longer to come on and then 9 seconds to
> manually slightly rotate a very small matrix 25,50 around.
>  So slow as to render the plotting function as useless as
> with the old mesh program.
Try "pkg load jhandles" or "toggle_octplot" and then run the graphics..

>   
> 5) On WinXP upon exit:  typing exit at the prompt works
> great.  Using the X in the corner to close the application
> keeps generating a 'crash' [the program fails] and WinXP
> keeps wanting to send reports to MS.  
>   
Known problem....

> Has anyone done benchmarks between these two?  It's just
> that most of my scripts take anywhere from 10 seconds up to
> 2 minutes to operate any slower is a catastrophe.  Worse I
> realy need great plotting in 3d that I can manipulate with
> the mouse - rotate and contemplate, etc.
Yes, there are benchmarks.. The ones for this 2.9.12-3 release had a few
issues in that I believe that ATLAS wasn';t used to compare against
2.1.50.. The speed is not catastrophic.. If your issue with your scripts
is purely the graphics then you should definitely try jhandles or octplot..


>   
>
> Wish I could contribute directly here.  But alas, I can
> only help by critiquing the 'way' it functions.  So, again,
> is there a place, or someone to communicate directly to, to
> discuss my observations?  For example, I have a 150KB
> screen shot comparing the script window for 2.1.50a with
> the one that is generated by 2.9.12.  
>   

2.1.50 uses rxvt from cygwin.. There are many reasons not to use rxvt
for 2.9.12 MSVC build.. It makes no sense to use an rxvt from cygwin
with an MSVC build of Octave, as you essentially have to include all of
the support libraries of cygwin just for the console.. As from the MinGW
build of rxvt, the pty emulation is seriously broken and this completely
messes up the use of readline with Octave, and so the prompt essentially
becomes unusable. Michael experimented with the console2 program (its
still in the installer but not chosen by default),  however the font
rendering of console2 is extremely slow, even if the console itself is
quite nice..

I expect that someone in the near future will wrap Octave on windows in
a full IDE and so I see no point spending too much time working on this
issue directly..

D.


> Robert
> _______________________________________________
> Help-octave mailing list
> address@hidden
> https://www.cae.wisc.edu/mailman/listinfo/help-octave
>
>   


-- 
David Bateman                                address@hidden
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 6 72 01 06 33 (Mob) 
91193 Gif-Sur-Yvette FRANCE                  +33 1 69 35 77 01 (Fax) 

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary



reply via email to

[Prev in Thread] Current Thread [Next in Thread]