Hello Vincent, hello Brice, > On 29/05/2014 12:14, Brice Videau wrote: > > Well this is normal. This is part of the process we used to retro > > engineer the loader when sources were not available. > > Nonetheless maybe we could let this part out of the main build. > > I'll let Vincent decide. > > I would keep it in the mail build. If I recall correctly, this code > is here to test the *ICD Loader* itself (with a specific *fake ICD*). > That explain why there is no need to stick to the real prototypes : > the first argument must be a pointer (required by the spec), other > arguments are not used by the ICD Loader. > Using fake (simpler) prototypes allow to generate automatic tests. > > Of course, using run_dummy_icd_through_our_ICDL with a real ICD would > probably crash. > > So, for now, my position is that I will keep this test in the > Debian package and I won't patch upstream software to tweak each > prototype manually (instead of using upstream automatic code > generation) in this test. > That said, if upstream find (or someone provides them) a way > to use the real prototypes, I will of course use them in the > Debian package. Or another possibility would be for you to tell > us how to desactivate this compiler check here (a pragma?) > [...]
Thank you very much for the detailed elaboration. I will take a closer look at the code to see what I can do to convince my tool that this is actually safe. I will get back to you once I've got that information. Best, Michael
pgpp4CcGD9Jwv.pgp
Description: PGP signature
