Re: [gmime-devel] ready to release 2.4.0, any last minute comments?

[Jeffrey Stedfast <fejj novell com>]

On Thu, 2008-09-18 at 20:13 +0300, djcb wrote:
> Hi Jeff,
> 
> On Wed, 17 Sep 2008, Jeffrey Stedfast wrote:
> 
> > I think I'm now ready to release GMime 2.4.0 as I'm pretty happy with
> > it.
> > 
> > I'll probably be rolling the tarball this weekend if there aren't any
> > objections, last minute requirements, suggestions, etc.
> > 
> > If anyone needs a little more time to review, just let me know - I'm not
> > in any rush ;)
> 
> Ah - nice! Also nice to see the recursive g_mime_multipart_foreach :-) 

:)

Yea, I implemented this after you commented about that the other day.
After thinking about it a bit, I came to the conclusion that it made no
sense not to have it be recursive :)

This mailing-list is already paying off ;-)

> 
> Couple of small things (based on SVN HEAD). 
> 
> * some compiler warnings on amd64/gcc4.3; seems rather harmless,
>   shouldn't stop the release I suppose.
> 
> * on my ubuntu/intrepid, the autogen.sh only works after:
> 
> 	case $libtool_version in
> -	    1.4*|1.5*)
> +	    1.4*|1.5*|2.2*)

ah, thanks - I'll have a fix committed in a sec.

> 
> * it seems gmime does not like the header encodings that Gmail produces
>   for 8859-1 special characters; ie. things like:
> 
>      From: =?ISO-8859-1?Q?Nordl=F6w_ <per nordlow example com>
>  
>   I guess g_mime_message_get_sender should turn that into UTF8 or
>   something? Instead I get: 
>   	     "=?ISO-8859-1?Q?Nordl=F6w_" <per nordlow example com>
>    
>   this happens even when compiling with -enable-rfc2047-workarounds

this is a bug in Google Mail's encoder which they are supposedly working
on fixing. It's breaking a lot of clients from what I understand.

I take it this is via IMAP?

> 
> * finally, i noticed that test-parser gives some GLib warnings due to
>   g_object_unref on some none-GObjects.

I bet it's because you don't have the mbox input files for the unit-test
and my unit-test code isn't properly handling failure to open them :p

I should commit the non-personal mbox files to svn and bundle them with
the gmime-2.4 tarballs - I've been meaning to do this for a while now,
anyway as it would help contributors to more-thoroughly test for
regressions.

I'll also fix the unit-test to better handle that case ;-)

Thanks for the input,

Jeff