Akim Demaille akim@epita.fr writes:
"Akim" == Akim Demaille akim@epita.fr writes:
"Giovanni" == Giovanni Palma giovanni@lrde.epita.fr writes:
Index: integre/ChangeLog from Giovanni Palma giovanni@lrde.epita.fr
- ntg/core/macros.hh: Add ntg_nbits macro.
Index: olena/ChangeLog from Giovanni Palma giovanni@lrde.epita.fr
- oln/utils/buffer.hxx: Add file.
- oln/utils/buffer.hh: Likewise.
- oln/utils/key.hh: Likewise.
- oln/utils/key.hxx: Likewise.
- oln/utils/md5.hh: Likewise.
- oln/utils/md5.hxx: Likewise.
I believed we had agreed this was not needed. Because if it were, then the code was already available and there was no need to spend time implementing again.
I am still worried with these changes. Just a few random thoughts, some of which might even not make sense.
- are you robust to the sex of the machines? (endianness)
I will check, but if integre works, md5 should works.
- you do save the pictures in files, don't you? So that someone who is facing a error can at least see the incorrect output.
No, he has to do it by his self, or to look at the doc where outputs can be seen...
- if it is not the case, then you just have to change the way you generate the tests to save the output file, that's a one liner change. *If* the tests are generated.
I aggree.
- if you save the files, then using md5sum is probably a good means to compute it.
Of course, but those images are computed with the doc and not with the tests.
To conclude, if someone is interested in producing a script to make such a work (generate tests with md5 signatures), I can give him the scripts i have written.