[poppler] [REMINDER] Poppler 0.6 release

Sat Aug 25 07:00:09 PDT 2007

El sáb, 25-08-2007 a las 11:50 +0200, Albert Astals Cid escribió:
> We planned 1st september, but i'm not going to be around that weekend so we 
> either release 30 august (5 days from now) or 3rd september (9 days from now)
> 
> I vote from 3rd september.

Ok to me. 

> I know Carlos wanted to introduce changes to glib and Pino to qt. Send them 
> NOW!

yeah, I'll do it the next week. 

> Also does anyone if Julien is around, we need him to have a look at 
> http://bugs.freedesktop.org/show_bug.cgi?id=12013 and say if that is needed 
> or if we can just remove the declaration from the .h
> 
> Also there are two things i want to discuss:
> 
> *** "unwrapping" of ff,fi, etc ligatures ***
> I don't want to do it, but that's only me, there seems to be lots of votes on 
> the other direction, what do you, Carlos, Jeff, Pino, krh say? And any of the 
> yes voters will send a patch? ;-)

Honestly, I don't have a strong opinion about this. 

> 
> *** beign more lax about ending %%EOF ***
> Lately there seem to be some pdf around that do not end with %%EOF. That sucks 
> as they break the check i introduced for %%EOF beign present and are not REAL 
> pdf.
> This check is not really needed for "finished" files, but on kpdf/okular we 
> have that handy "Check for file changes" that reloads the document once the 
> file has changed (useful for latex people), the problem with that is that you 
> know it changed, but not that it will not change anymore, so we often ended 
> up reloading incomplete files that randomly crashed poppler.
> So i introduced the %%EOF check to avoid that.
> But it seems that is hurting us now, so we can remove the %%EOF check and try 
> to improve poppler not to crash on incomplete documents or leave it that way 
> and keep repeating our users that if it does not have an %%EOF it's not a PDF 
> and that they should bug the producer not the consumer of the pdf file.
> Or any other suggestion?

I wouldn't like okular/evince/whatever-viewer-depending-on-poppler users
have to use acroread just because poppler crashes when acroread not. 

I would like to add another two things to discuss:

*** ZapfDingbats fonts ***

These fonts are used in documents with forms to display checkboxes and
radio buttons symbols. Right now, fontconfig doesn't find the right font
when asking for 'ZapfDingbats', which means that documents with forms
(if they have checkboxes or radio buttons) don't work for most of the
users. And it fails silently, when user clicks on a radio button or
checkbox nothing happens. It seems that the problem should be fixed in
every distro, so if we fix it in poppler it will never be fixed by
distros. Actually I don't mind how is to blame, the truth is that
interactive forms is one of the most important features of the new
evince version and I'm sure it's very important in okular too, and it's
currently broken. I tried to do the right thing so I commented it in
fontconfig bugzilla[1] and filed a bug[2] to my distro (ubuntu) with no
responses in both cases. My proposal for poppler 0.6 is try to figure
out whether fontconfig will be able to find the right font with some
"magic" in the configure script (using fc-match command). If it fails,
we show a warning indicating that there is a problem in the fontconfig
configuration that should be fixed, and we define a preprocessor macro
indicating that the substitution has to be done by poppler (as a
workaround). The changes in poppler are minimal, something like this:

#ifdef SUBSTITUE_ZAPFDINGBATS
if (strcmp (family, "ZapfDingbats") == 0) {
  family ="Dingbats";                                                                                                                                      
}
#endif

*** Remove splash dependency in glib bindings ***

In other words, make glib bindings strictly depend on CairoOutputDev and
remove all the #ifdef crap. This is something krh proposed for 0.6
which, at this moment, I agree with. AFAIK most of the distros are
already compiling poppler with cairo outputdev enabled, so the glib
bindings are using cairo. 

[1] https://bugs.freedesktop.org/show_bug.cgi?id=6478#c12
[2] https://bugs.launchpad.net/ubuntu/+source/fontconfig/+bug/130687

-- 
Carlos Garcia Campos
   elkalmail at yahoo.es
   carlosgc at gnome.org
   http://carlosgc.linups.org
PGP key: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x523E6462
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Esta parte del mensaje =?ISO-8859-1?Q?est=E1?= firmada
	digitalmente
Url : http://lists.freedesktop.org/archives/poppler/attachments/20070825/97b312a9/attachment.pgp 