Re: guile-gdbm doesn't work with gdbm-1.14

[Mark H Weaver <mhw@netris.org>]

Ricardo Wurmus <rekado@elephly.net> writes:

> Mark H Weaver <mhw@netris.org> writes:
>
>> iyzsong@member.fsf.org (宋文武) writes:
>>
>>> Hello, since version 1.14, gdbm doesn't export "gdbm_errno" anymore [1],
>>> so the guile-gdbm ffi binding code [2] need updates now (I'm not
>>> confident to do it myself...).
>>>
>>> [1] http://git.gnu.org.ua/cgit/gdbm.git/commit/?id=c175231e2781abd17eabf412cfb597654a076c7b
>>> [2] https://github.com/ijp/guile-gdbm/blob/master/gdbm.scm#L156
>>
>> Here's a preliminary fix.
>
> Thank you.
>
>> * gnu/packages/patches/guile-gdbm-ffi-support-gdbm-1.14.patch: New file.
>> * gnu/local.mk (dist_patch_DATA): Add it.
>> * gnu/packages/guile.scm (guile-gdbm-ffi)[native-inputs]: New field.
>> [inputs]: Move above arguments.  Add the patch, and the 'patch' program.
>> [propagated-inputs]: Move above arguments.
>> [arguments]: In the builder, add code to apply the patch.
>
> I don’t see the native-inputs field in the patch.

Indeed, sorry for the mistake in the commit log.  Initially I made them
native inputs, but then I moved them to 'inputs'.

> Shouldn’t the “patch” and “patch-file” inputs be native-inputs?

Yes.  However, I noticed that the package already assumes a native
build, because it runs 'guile' from 'inputs' to compile the Scheme code.
Also, I wasn't sure off-hand how native-inputs are handled in the
trivial-build-system.

For purposes of this commit, I didn't want to take on the job of also
fixing this package for cross-building, which I was likely to get wrong
without testing.  Given that I no longer use substitutes, that would
have been a big job.

Does that make sense?

      Mark