> I took a look at this and it's very cool! Some really nice things in > here. Kudos!
I agree. It's a nice start. I only wish I had more time at this point to contribute :( I'll get some commentary going on gmsl, but perhaps it should go to a different mailing list? >From an implementation standpoint, is the desire of gmsl to stay out of the >$(shell ...) environment for all/most/some of its implementation? For >example, my trace mechanism does not use warning because I really don't want >makefile to spit that information out. I use shell and echo and redirect to >the stderr. Since there appears to be no make function to echo information >without adding information to it, I've gone with doing it this way, although >I'd prefer making a patch to gmake ;). There are some other things which I use sed as a text formatter for some of my functions. Could there be a supported list of shell functions that would be needed to implement some of gmsl's functionality? I currently use echo, sed, rm, and test, although I try to limit it to that, and also try to keep as much functionality in make as possible as ther is a performance penalty. Also, it will probably happen at some point that the maintenance of gmsl will become burdensome if all the implementation is in a single file. Eventually, I think there will come a point where the implementation needs to be split into several files. It does have the advantage that there aren't many function dependency issues though. Lastly, a note on syntax. gmake has functions which use '-' instead of '_' to separate words in functions (i.e. filter-out), so should this continue in that tradition? My functions currently use '_' as well, but I think maintaining some uniformity with gmake's syntax is something to consider. Best, John _______________________________________________ Help-make mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-make
