Hi Bruno, Le vendredi 08 août 2008 à 00:01 +0200, Bruno Haible a écrit : > > I'm not competent to write the Win32 implementation unfortunately. Will > > you be able to do this part of the work? > > Yes. Win32 has the necessary primitives, it's only a matter of looking them > up... I can do that. > > > > The prefix 'gl' for the module and header file is probably unnecessary: > > > most modules are gnulib modules anyway, and if the header was called > > > "cond.h", > > > there would not be much risk of collision either. > > > > From experience, I'd prefer to avoid possible collision. Not that I ever > > met a system defining cond.h, but rather that I find it to be a good > > habit to have a generic prefix for this kind of stuff. I prefer > > preventing potential issue than curing them afterwise. > > Is that a problem to you? > > It's not a big problem. But when we have a family of 4 modules, it would be > strange if they would use different conventions for the header files. - I used > the prefix 'gl_' in the ordered lists and sets modules, because I did not > dare to call a header file "list.h". But "cond.h" should be fine. For your > 4th module, however, "thread.h" would collide with the Solaris <thread.h>.
Since we have to keep glthread anyway, doesn't it make sense to use glcond / gllock to use the same convention everywhere? -- Yoann Vandoorselaere <[EMAIL PROTECTED]>
