On Mon, 2008-01-14 at 08:56 -0500, Matthias Clasen wrote: > 2008/1/14 Owen Taylor <otaylor redhat com>: > > > > Comments, criticisms? > > > > I'm slightly concerned by this ... libgweather does something pretty > > interesting (to a small set of consumers) but the API is no way ready > > for any sort of freezing. It has serious namespace and memory management > > issues, among other things. > > > > I don't think there is any proposal here to make libgweather part of the > > platform, but a) is there clear messaging about the API status? b) how > > would a transition to an incompatible new version be handled? > > Pretty much like for any other instable api, I think. > Users will have to check for a version they support and will have to be ported. > > Do you want us to do the -DI_SWEAR_I_KNOW_LIBGWEATHER_IS_UNSTABLE dance ? > That is easy enough, if you think it helps. While that dance seems annoyingly pointless at times, I think it does serve some small purposes: - Everybody who compiles the modules using the unstable API sees it repeatedly and it hopefully sinks into the unconscious - It can be pointed to when people start complaining after an API break. - Owen
Attachment:
signature.asc
Description: This is a digitally signed message part