Refactoring up-imapproxy

Tim Jackson lists at timj.co.uk
Sun Jan 6 11:36:38 UTC 2008


Thorsten Leemhuis wrote:

[inconsistent naming of stuff in up-imapproxy]

> * there is no rule that EPEL packages have to be similar to the one from
> Fedora or based on those from Fedora (maybe such a rule should exist to
> prevent users bypass review), but it helps everyone a lot afaics; fixes
> can easily be floating back and forth and (more important) users that
> know the package from Fedora can just use the one from EPEL (or vice
> versa); that might not sound very important, but it IMHO is as
> differences like those outline above create a lot of confusion for users

Indeed, which is why I wanted to involve other people. In a way it's a 
choice of the lesser of two evils: inconsistencies between Fedora and 
EPEL, and inconsistencies within the package itself.

> * what upstream does here looks to be confusing, but fixing this
> confusion only in Fedora-land (either Fedora or EPEL or both) just adds
> more confusion for everyone, as Howtos and Docs written from upstream or
> written for other distributions won't "just work" in Fedora-land as
> files are stored in different places

Fair point.

> Thus is might be better for everyone to stick to the package design
> what we have in Fedora for now, fix the naming problem upstream and then
> adjust the packaging in Fedora-land to it.

OK, I think that was a well-reasoned argument. However, one thing I 
omitted to mention was that the current layout does cause the following 
from rpmlint:

W: incoherent-init-script-name imapproxy

I do think if we are going to call the package "up-imapproxy" then the 
init script should also be "up-imapproxy" (even if we leave the config 
file location alone), and this is something that we are not "fixing" in 
Fedora land but rather just being self-consistent in doing. This is 
probably the thing that annoys me most, since I inevitably (in Fedora) 
try to do "service up-imapproxy start", having just installed a package 
called "up-imapproxy". I personally thing this is a mistake in the 
Fedora package and is worth fixing for the EPEL branch (and probably in 
Fedora for F9 too, although the current Fedora maintainer seems to be 
unresponsive).

Tim




More information about the epel-devel-list mailing list