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

Re: Maintainers, porters, and burden of porting

* Lucas Nussbaum <lucas@lucas-nussbaum.net> [110829 19:28]:
> I think that bugs caused by important differences compared to other
> Debian architectures are the porter's job to handle, not the
> maintainer's.
> That includes stuff like:
> - missing/different packages on $ARCH
> - missing/different libraries on $ARCH
> - different libc semantics on $ARCH (hi kfreebsd+hurd!)
> etc.

But I think it is definitely the maintainer's job to handle bugs in
his source package including (but not limited to):

- broken alignment
- broken handling of endianess
- source code with undefined behaviour only currently having visible
  effects on one/some architectures.
- different libc implementation

Those are bugs in the package and the "differenced compared to other
Debian architectures" are only working as a way to catch those bugs
before they hurt even those using more 'mainstream' architectures.

Assume for example the latest "memcpy copies backwards" change
in libc. If that had been only been implemented on some specific
non-mainstream architecture, that would not have been a job for
porters to care about, but for maintainers, as it is clearly a
bug in a package to assume one can use memcpy where memmove is to
be used.

(And try to imagine how hard it would have been to introduce amd64
if alpha had not elliminated in many years work most of the subtle
64 bit bugs found in most software, I doubt porters alone could have
completed this in that time).

	Bernhard R. Link

Reply to: