bug#1524: 23.0.60; gs.el - This code is experimental. Don't use it.

[Leo <sdl.web@gmail.com>]

On 2009-01-11 14:57 +0000, Stefan Monnier wrote:
[...]
> IIRC David Kastrup would know how broken it is.

Here is David's response.

------------------------------------------------
Leo <sdl.web@googlemail.com> writes:

> 2009/1/11 Stefan Monnier <monnier@iro.umontreal.ca>:
>>>>> I think it is safe to remove (info "(elisp)PostScript Images") and
>>>>> gs.el now that over a month has passed.
>>>> It's used in image.c.  We should probably try to fix it if it
>doesn't
>>>> work properly.
>>> I don't use it so I don't know whether it has bugs. I suspect it
>>> has. But I'm mostly concerned by the comment in the source which
>advise
>>> users not to use it.
>>
>> IIRC David Kastrup would know how broken it is.
>
> Stefan's email sent an email address of yours which is no longer in
> use. Could you comment this in emacs-devel mailing list?

Pass it on if you want to.  There are a lot of problems due to
Ghostscript rendering directly to Pixmaps.  Every image is rendered with
its own Ghostscript process, but the parameter passing requires X
properties as a global resource.  This is done once per Ghostscript
session and includes the display dimensions, so it can't really be
extended to multiple image rendering without further Ghostscript API
changes.

The display code triggering the display is rather asynchronous: I've not
been able figuring out how it is rooted in the bowels of the display
engine.

In practice, it worked more or less for a single image rendering per
Emacs session.  It doesn't scale at all.  Serious work both in
Ghostscript as well as in Emacs would be required to make this a smooth
working feature, and I see no candidate for either.
------------------------------------------------