[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: MinGW

Callum Lerwick wrote:
> On Wed, 2008-09-10 at 16:02 +0100, Richard W.M. Jones wrote:
>> On Wed, Sep 10, 2008 at 10:03:07AM -0400, Josh Boyer wrote:
>>> Can you comment on what part of his draft you find objectionable?
>> Specifically three things:
>> (1) It imposes upon us the need to use a separate repository, which is
>> based on the false assumption that we will be rebuilding a substantial
>> proportion of all Fedora packages, like some sort of secondary
>> architecture.
>> In reality this is not the case - we only wish to rebuild a few common
>> libraries.  Secondary architectures rebuild every package, including
>> applications, which we have no intention of doing even if it were
>> possible (which it isn't).
> I think all this arguing about "number of packages" is a red herring.
> This all smacks of misguided political compromise where pointless
> obstructions are put in place in an attempt to appeal to both sides.
> This kind of bullshit has no place in an engineering organization. If we
> could point to a repo overflowing with mingw packages, there might be a
> point to this. But right now it's purely "oh there might be a lot of
> them!" which is not sound engineering practice, it's hypothetical
> bullshitty handwaving.
> Either we commit 100% to having a Win32 toolchain in Fedora or we don't
> bother. No compromises.
I agree with your sentiment.  I'm not certain that a separate repo is a
bad thing, though... it allows mirrors to choose whether to mirror the
content or not.  If we're committing 100% to to having a win32
toolchain, though, we should enable the repo by default.

>> (2) "All packages must first be natively available in Fedora before
>> they can be in the MinGW repo"
>> This is a considerable restriction.  A useful Windows cross-
>> development environment must include packages like NSIS installer, GNU
>> gettext and PortableXDR, none of which would make sense as standalone
>> Fedora packages.
> It's another pointless obstruction. If a cross-package meets licensing
> guidelines, why is it not be acceptable?
This is one that I'm not so sure about.  The theory is that there's
packages which belong in a cross-compiled environment but not in Fedora.
   But the question is what is the definition of "belongs"?  NSIS is
useful for Fedora and Richard has patches to let it work to some extent
here.  gettext *is* in Fedora for the applications that are in the
package.  PortableXDR is a library that would provide functionality
redundant with stuff in glibc but on the other hand we have tons of
packages which provide redundant functionality in Fedora already.

If someone proposed these packages be natively compiled and entered into
Fedora we would welcome them.  So the argument that these don't make
sense as standalone packages in Fedora is not quite true.

On the other hand, what we want to know is why we should impose an
additional restriction on packages destined for use in a crosscompiler.
   Jef, care to lay out the reasons for this restriction?


Attachment: signature.asc
Description: OpenPGP digital signature

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]