This week had a continuation on the discussion regarding improving debug info at various optimization levels and the general consensus was that behaviour that caused code generation differences with and without using -g should be removed. This is a basic promise by GCC to its users and would result in a lot of things breaking in such a situation. It would be interesting to wait and watch on this space since it promises to improve things with debug info as well as try and define compromises with debug info and optimizations (as long as optimizers do not generate code dependent on debug info being turned on are not .) .
Dan Berlin started a huge thread about using git for the version control system and the space / time overheads with checking out a repository , keeping patches and generally using git instead of svn and / or mercurial. There is a huge thread that starts here and goes on to include the friendly neighbourhood folks from the git community in attempting to get git working on the gcc community. I've tried in the past using git with an internal CVS repository and given that I work only with relatively small history I haven't had the kinds of issues that folks are having on the lists.
However the scales are different and it would be interesting to see what is the kind of times I get using Mercurial to play with. Makes sense for a weekend project.
BTW the command line for Mercurial is hg, in some sense the chemical formula for Mercury .:) .
No comments:
Post a Comment