Reading the article and related I find it interesting that for:
* tiny compile perf increase (as per Michaels benchmarks)
* Somewhat better error mesages (but, really GCC 8 stepped up a lot, the difference is getting smaller)
* Can run git snapshots
They have to:
* run on git snapshots to get bugfixes
* handle thousands of patches
* many of those patches are disputed as having to work around deficiencies in llvm (which turns this into a political game)
* and hundreds more packages in your backup compiler in any case
* support less architectures, so in some arch you revert back to your old compiler.
This is called "great success"…
This made me pause.
The only reason you would do this is I can think of is for money, academic or as part of a war on "justice".
Yes, I get it GCC needed a kick in the butt to get moving, but, seriously, have you seen the constant improvements since gcc 4.6 onwards? I sure have.
* tiny compile perf increase (as per Michaels benchmarks)
* Somewhat better error mesages (but, really GCC 8 stepped up a lot, the difference is getting smaller)
* Can run git snapshots
They have to:
* run on git snapshots to get bugfixes
* handle thousands of patches
* many of those patches are disputed as having to work around deficiencies in llvm (which turns this into a political game)
* and hundreds more packages in your backup compiler in any case
* support less architectures, so in some arch you revert back to your old compiler.
This is called "great success"…
This made me pause.
The only reason you would do this is I can think of is for money, academic or as part of a war on "justice".
Yes, I get it GCC needed a kick in the butt to get moving, but, seriously, have you seen the constant improvements since gcc 4.6 onwards? I sure have.
Comment