[OT] who maintains aptitude for rpm? (fwd)

[Panu Matilainen]

Sorry, forgot to CC you initially...

 	- Panu -

---------- Forwarded message ----------
Date: Tue, 3 Oct 2006 22:34:14 +0300 (EEST)
From: Panu Matilainen <pmatilai at laiskiainen.org>
To: Discussion about apt-rpm <apt-rpm at lists.laiskiainen.org>
Subject: Re: [OT] who maintains aptitude for rpm?

On Thu, 14 Sep 2006, Wilfried Weissmann wrote:

> Hello,
> 
> Sorry for posting something offtopic but this is the best place I could find 
> to ask this question. Who maintains aptitude for rpm? The latest verion that 
> I could find is 0.3.3 from Mandriva, however there is no project page.

I don't think there's such a thing as maintainer for aptitude on rpm, and I'm 
not too interested in becoming one :) What I can certainly do is collect 
relevant patches to apt-rpm.org so people have a chance of finding them.

> I am asking because I have two problems with aptitude. The first is that the 
> package description is parsed by aptitude and rendered in a way that is only 
> suitable for DEB packages. This parsing breaks the description of RPMs which 
> have "raw" text in the description field. I have attached a patch that just 
> disables the parsing and just copies the discription line by line. The other 
> issue is that if I uninstall a package that has a %postun scriptlet then the 
> script always fails without being executed (!) and when aptitude returns to 
> the main screen it crashes with SIGSEGV.
> 
> Committing changes...
> Preparing...                ########################################### 
> [100%]
> error: %postun(blahlaber-1-0.noarch) scriptlet failed, exit status 255
> Press return to continue.
> 
> Ouch!  Got SIGSEGV, dying..
> Segmentation fault
> 
> If I remove the package with "apt-get remove blahlaber" (version 
> apt-0.5.15lorg3.2) everything works fine. This looks to me like there is some 
> memory corruption in the aptitude code.

What distribution (and version) are you using, and do you by any chance have 
SELinux enabled? Scriptlets failing in aptitude but working in apt *could* be 
explainable with SElinux context/type differences. The fact that aptitude 
crashes because of that is obviously a bug of some sorts - one possibility 
might be that it blindly assumes a package marked for removal actually getting 
removed no matter what (if scriptlets fail rpm's dont get removed!) and gets 
its bookkeeping wrong because of that (or something like that). But that's just 
wild wild speculation, haven't actually looked at aptitude code...

 	- Panu -