[argyllcms] Re: Argyllcms 1.0.1 packaged in fedora-devel

On Sun, Jul 27, 2008 at 4:47 PM, Graeme Gill <graeme@xxxxxxxxxxxxx> wrote:
> Nicolas Mailhot wrote:

>>
>> http://cvs.fedoraproject.org/viewcvs/devel/argyllcms/argyllcms-1.0.1-remove-libusb-fork-check.patch
>>  Why oh why do we have to do this? I Notice Mandriva had to do
>>  something similar. As we said before if there's a specific problem in
>>  libusb report it and we'll get it fixed
>
> It's to make sure that Argyll works. If you're not using the provided
> libusb,
> it almost certainly won't work with certain instruments. While I attempted
> to
> get upstream to change libusb to fix the threading problems that crop up
> with
> the i1pro, I don't think they have done so or have any plans to do
> so since they want to "fix" it in the next, unfinished, incompatible
> libusb V1.0. So unless you are prepared to patch your libusb and test
> it with all of the supported instruments, please don't repackage
> Argyll in a form that is probably broken. How do you propose
> to fix the bug you've introduced in your version of Argyll ?

Well, who said we introduced a bug in Argyllcms ?

I made sure to contact people who had issue with your first patch on
libusb, when I tried to integrate it in cooker some months ago.
Fortunately, those issue were corrected so I integrated your patch in
our cooker libusb and I had to disable your ugly code check in
argyllcms (which, btw, is wrong, since unpatched argyllcms running on
a  non-argyllmcsified libusb will not even be started, since a symbol
will be missing (and you won't see the "expected" error message).

I still think it is extremely ugly to do this kind of hack, without
any previous warning to distributions like us. I discovered this while
packaging argyllcms 1.0.0 and manually diffing your patched copy of
libusb with a "upstream" libusb.

-- 
Frederic Crozat

Other related posts: