On Wed, 19 Apr 2006, Heikki Toivonen wrote:
Andi Vajda wrote:
My preference is neither. We should document which compilers are known
to work so that developers who want reliability know what to work with.
I don't see how this could work. Any code change could cause the build
to break under some previously "known to work" compiler.
Yet it works great today. We know quite well which compilers work on the three
OSs we support. Our build is using them. While any code change could cause
lots of problems, usually that is not the case. And that is a good thing.
If a particular compiler, say gcc 4.x on Linux or Mac OS X, is particularly
sensitive to some wx code changing for example, we'd consider taking it off
the compilers-known-to-work list. We might even add it to the list of
compilers-known-not-to-work once we know what code is causing what bug in the
compiler. We should also consider filing that bug, most compilers we use are
open source, after all. In other words, open source compilers get better
because people use them and find bugs that get fixed.
Andi..
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Open Source Applications Foundation "chandler-dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/chandler-dev
