gcc - OUT; clang - IN?

[scsijon]

Rob Landley (Aboriginal Linux and Toybox) and a few other 'Basic Linux System' Groups are talking at present and considering replacing gcc with clang (as I read their messages).

I wonder if one of our bleeding edge people would consider looking into the basics of a similar Puppy conversion for us and reporting back on the feasabilities and workload?

thanks

scsijon

.b typos, sorry I need a new kbd

[jamesbond]

Before any effort is expended, the first question to ask is: why? what's the benefit? what clang can do that gcc can't?

[amigo]

clang can't do (still) some of what gcc does.

[scsijon]

I didn't ask for a detailed report, just a basic look-at, comparison and report back possibly by a Puppy WOOF-CE System 'techie / programmer' or an interested party! It shouldn't take forever.

Clang main page is http://clang.llvm.org/ and comparison page is via http://clang.llvm.org/comparison.html for those interested.

The only problem I see as a possible stop to start with is the LLVM 'BSD' License, but is that now a problem with all the licences the packages and systems puppy are now buiit against?

[jamesbond]

I read your OP as asking for "report for performing the conversion". This, plus the "provocative" title of "gcc is out, clang is in" gives me the impression that you're already convinced that replacing gcc is a good thing, and the only thing left to consider is the effort.

I'd like to step back a little and ask you to explain the reason behind your conviction (other than "Rob Landley is considering that too...").

PS: I have a lot of respect for Rob. The tool he create (Aboriginal Linux) is one of my key build tools. But that doesn't mean I always agree with his opinions.

[scsijon]

I read your OP as asking for "report for performing the conversion". This, plus the "provocative" title of "gcc is out, clang is in" gives me the impression that you're already convinced that replacing gcc is a good thing, and the only thing left to consider is the effort.

This is the cutting edge forum to start with so the title is appropriate for the topic.

I provided two lines of topic, the first one contained why I was interested and the second one stating what I was asking someone with enough 'nouse' and some interest to do so as i'm too ill at present (which is why all my projects both puppy and non-puppy are in a halt stage) to attempt it and it might at some stage give us an 'edge' again as we had in the past when puppy 2.13 was a standard!

To quote myself, I asked for the following to happen if someone was interested.

I wonder if one of our bleeding edge people would consider looking into the basics of a similar Puppy conversion for us and reporting back on the feasabilities and workload?

I also did not ask for any such thing as actually doing the conversion for all, nor did I sugest we should change, just the feasability of creating a puppy of this build type.

I have no idea on the outcome so don't put words in my mouth please, and if your not interested in actually doing the project, please don't fill the thread with statements not related to what is asked for.

e.o.s.!

regards to others
scsijon

[jamesbond]

Not attempting to put words in your mouth at all. If my post came sounding like that, then my apology.

I was just trying to point out that a puppy build from the ground-up with clang will have absolutely no visible difference with puppy build from gcc. That's why I keep asking you why you think it's a good idea to replace it. clang has a bit of edge in compiling speed, but gcc has an edge that the binary it produces usually run faster. But people who prefer clang can always install clang package for their compiling needs; gcc and clang can be installed side-by-side, no problem.

In my opinion, the choice between clang and gcc these days generally revolves around which license you like best, BSD/MIT (clang) or GPLv3 (gcc).