23.1.95 pretest; windows binaries

23.1.95 pretest; windows binaries

[Sean Sieger]

The current pretest is available now at

http://alpha.gnu.org/gnu/emacs/pretest/windows/

RE: 23.1.95 pretest; windows binaries

[Drew Adams]

> The current pretest is available now at
> http://alpha.gnu.org/gnu/emacs/pretest/windows/

Thanks, Sean.

Re: 23.1.95 pretest; windows binaries

[Christoph]

On 4/5/2010 10:35 AM, Sean Sieger wrote:
> The current pretest is available now at
>
> http://alpha.gnu.org/gnu/emacs/pretest/windows/
>    
Thanks Sean.

Two questions:

1. This pretest, like the last one, contains more DLLs than the usual 
libXpm.dll. Is this intentional?

2. Did you remember to execute 'make info' before packaging? The size of 
this current pretest differs by about 1.5 MB from the 23.1.94 and upon 
further investigation it turns out the info directory does not contain 
any of the files produced by makeinfo.

Christoph

Re: 23.1.95 pretest; windows binaries

[Jason Rumney]

> 2. Did you remember to execute 'make info' before packaging? The size 
> of this current pretest differs by about 1.5 MB from the 23.1.94 and 
> upon further investigation it turns out the info directory does not 
> contain any of the files produced by makeinfo.

The info files should have already been built in the source tarballs.

Re: 23.1.95 pretest; windows binaries

[Sean Sieger]

    The info files should have already been built in the source tarballs.

That's what I thought.

Re: 23.1.95 pretest; windows binaries

[Christoph]

On 4/5/2010 6:37 PM, Jason Rumney wrote:
>
>> 2. Did you remember to execute 'make info' before packaging? The size 
>> of this current pretest differs by about 1.5 MB from the 23.1.94 and 
>> upon further investigation it turns out the info directory does not 
>> contain any of the files produced by makeinfo.
>
> The info files should have already been built in the source tarballs.
>
Hmm...true. The info files are indeed in the tarball. That's weird.

Re: 23.1.95 pretest; windows binaries

[Sean Sieger]

    1. This pretest, like the last one, contains more DLLs than the usual
    libXpm.dll. Is this intentional?

Do they not need to be in /bin to support the image formats that Emacs
supports?

    2. Did you remember to execute 'make info' before packaging? The size
    of this current pretest differs by about 1.5 MB from the 23.1.94 and
    upon further investigation it turns out the info directory does not
    contain any of the files produced by makeinfo.

make info returns something like `Nothing to make' as the tarball
that I build from already has a complete set of info files.

I've just started looking at the differences in the sizes of the those
zip files (23.1.91 51MB; 23.1.94 52MB; 23.1.95 48MB), I too am looking
forward to finding the cause.

Thank you, Christoph.

Re: 23.1.95 pretest; windows binaries

[Jason Rumney]

On 06/04/2010 10:23, Sean Sieger wrote:
>      1. This pretest, like the last one, contains more DLLs than the usual
>      libXpm.dll. Is this intentional?
>
> Do they not need to be in /bin to support the image formats that Emacs
> supports?
>    

They need to be somewhere on the PATH.  But Emacs doesn't require those 
image formats to display its own interface correctly (for example the 
splash screen and the tool bar), and if you include binaries of those 
libraries with Emacs, then you also need to include the exact source 
they were built from along with them (as is done in 
libxpm-3.5.7-w32-src.zip). I'd also advise to keep up with security 
bugfixes in those libraries so we are not shipping known security holes 
to Emacs users.  The last task is what convinced me to keep the 
libraries to a minimum, as at the time I was considering this there was 
a spate of security holes found in image libraries. Also, building some 
of these libraries on Windows is not supported by the standard makefiles 
and is poorly documented.

RE: 23.1.95 pretest; windows binaries

[Drew Adams]

FYI, just as for the 23.1.94 Windows binary, `report-emacs-bug' sends reports to
bug-gnu-emacs@gnu.org instead of emacs-pretest@gnu.org. Dunno whether this
really matters, but it doesn't seem right.

RE: 23.1.95 pretest; windows binaries

[Drew Adams]

> FYI, just as for the 23.1.94 Windows binary, 
> `report-emacs-bug' sends reports to
> bug-gnu-emacs@gnu.org instead of emacs-pretest@gnu.org. Dunno 
> whether this really matters, but it doesn't seem right.

I meant emacs-pretest-bug@gnu.org.

Re: 23.1.95 pretest; windows binaries

[Jason Rumney]

On 06/04/2010 13:47, Drew Adams wrote:
>> FYI, just as for the 23.1.94 Windows binary,
>> `report-emacs-bug' sends reports to
>> bug-gnu-emacs@gnu.org instead of emacs-pretest@gnu.org. Dunno
>> whether this really matters, but it doesn't seem right.
>>      
> I meant emacs-pretest-bug@gnu.org.
>    

I think it is correct. Since the bug tracker was installed, there is no 
longer a separate mailing list for pretest bugs.

Re: 23.1.95 pretest; windows binaries

[Sean Sieger]

Thank you, Jason.

Re: 23.1.95 pretest; windows binaries

[Sean Sieger]

Jason Rumney <jasonr@gnu.org> writes:

    They need to be somewhere on the PATH.  But Emacs doesn't require
    those image formats to display its own interface correctly (for
    example the splash screen and the tool bar), and if you include
    binaries of those libraries with Emacs, then you also need to include
    the exact source they were built from along with them (as is done in
    libxpm-3.5.7-w32-src.zip). I'd also advise to keep up with security
    bugfixes in those libraries so we are not shipping known security
    holes to Emacs users.  The last task is what convinced me to keep the
    libraries to a minimum, as at the time I was considering this there
    was a spate of security holes found in image libraries. Also, building
    some of these libraries on Windows is not supported by the standard
    makefiles and is poorly documented.

I pruned and built a libXpm-3.5.8 libXpm.dll and will upload it with the
next pretest or release; I removed all other image support dlls.  Will
look into complete set of currents image support libraries unless you
advise against spending my time doing so.

I think the libxpm sources could be pruned a little closer, I would
appreciate any help with it and maybe testing it with emacs-20100409-bin
& barebin-i386.zip, I will build a snapshot of the trunk tonight and
upload it to ftp.gnu.org/gnu/emacs/windows along with
libxpm-3.5.8-w32-src.zip.

Re: 23.1.95 pretest; windows binaries

[Richard Stallman]

    1. This pretest, like the last one, contains more DLLs than the usual 
    libXpm.dll. Is this intentional?

Does it really contain these DLLs, or does it only use them at run
time?

Re: 23.1.95 pretest; windows binaries

[Christoph]

On 4/6/2010 9:20 PM, Richard Stallman wrote:
>      1. This pretest, like the last one, contains more DLLs than the usual
>      libXpm.dll. Is this intentional?
>
> Does it really contain these DLLs, or does it only use them at run
> time?
>    
By "contain" I referred to the pretest package (zip file). Emacs uses 
these libraries at run time.

Re: 23.1.95 pretest; windows binaries

[Eli Zaretskii]

> From: Richard Stallman <rms@gnu.org>
> Date: Tue, 06 Apr 2010 23:20:51 -0400
> Cc: emacs-devel@gnu.org
> Reply-To: rms@gnu.org
> 
>     1. This pretest, like the last one, contains more DLLs than the usual 
>     libXpm.dll. Is this intentional?
> 
> Does it really contain these DLLs, or does it only use them at run
> time?

The pretest tarball (with the sources) does not contain the DLLs.  But
the binary distribution of the pretest for Windows does include the
DLLs.  They are needed for proper display of the tool-bar buttons.