On Tue, 2006-07-25 at 13:26 -0500, Rex Dieter wrote:
Ralf Corsepius wrote:
> On Tue, 2006-07-25 at 12:40 -0500, Rex Dieter wrote:
>> Ralf Corsepius wrote:
>>> On Tue, 2006-07-25 at 18:41 +0200, Axel Thimm wrote:
>>>
>>>> Let's file it under hear-say then and move on.
>>> No comment
>>>
>>> c.f. below and note the output of the "echos".
>> I think it's safe to just say this is an example of bad rpm practice.
>> If you really want/need two different sources and versions, package them
>> separately.
>
> ... you are ignoring the fact that there exist cases where this is
> impossible.
Seriously, it's bad practice, don't do it. But don't mind me, go ahead
and do it, if it's so "impossible" to do otherwise...
Check out any one tree style built GCC+newlib rpm,
check out autogen + libopts (currently under review).
just don't
complain when it doesn't work.
Bummer, a tool (here: rpm) isn't broken,
just because it doesn't fail on
the 90% of trivial cases it is being used by, but fails on the remaining
10% of complex cases?
Ralf