Re: Poor plot performance on Windows

[Mike Miller]

On Wed, Mar 16, 2016 at 12:07:57 -0700, Jake wrote:
> Very familiar with open-source development. Going to offend many by saying
> so, but here's the unavoidable truth about what makes a project good or bad:
> 
> - Bad: developers only work on what they personally need and/or are
> dismissive of others' issues. Those same developers tend not to help others
> get started because they only care about their own problems.
> - Good: developers work on their own problems + others' problems.
> Developers care about the quality and completeness of the project and work
> towards "finishing" the project, not just making it "good enough".

No offense taken here, I agree for the most part.

Personally I am interested in helping make Octave as stable and portable
as possible. I test and fix bugs on a wide range of GNU/Linux and UNIX
systems even though I personally only use Debian.

I want Octave to be just as useful and stable on Windows as it is on
Debian. I have no access to a Windows system to help make that happen,
so all I can do is make suggestions for others to try and ask for test
cases and debugger stack traces.

If someone is willing to give me remote access to a Windows system or
buy me a copy of Windows to install in a VM, it would be at least
possible for me to test and debug things directly.

> On principle alone, the plot window should not take more than a few seconds
> to appear, period. This problem has been around for a long time now,
> reported by multiple people multiple times. I have used Octave off and on
> for years and basically I can't use it more because my main need is also
> MATLAB does it faster and more consistently, therefore it is possible.

Back to your original post, it would still be helpful to see an example
demonstrating this behavior. Is the main delay in the on-screen plotting
or in the saving to a PNG file? Is saving to an EPS file faster? Does
the delay have to do with the complexity of the figure?

-- 
mike