IRC channel logs
2025-11-01.log
back to list of logs
<eyalroz>Here to check whether the GCC bootstrapping chain described here: <eyalroz>is still relevant today, or whether there have been further improvements/simpilifcations since. <ekaitz>eyalroz: we made some things on top of that <ekaitz>currently, we skipped over gcc 2.95 <eyalroz>ekaitz: Oh, I think the chain I saw doesn't have gcc 2.94... <ekaitz>eyalroz: we build 4.6.4 with tcc directly <ekaitz>in fact, i forked gcc and maintain a copy of 4.6.4 with riscv support <eyalroz>Do you suppose you (personally or bootstrappable.org) could perhaps post something more up-to-date about the bootstrap chain? It is both educational and of practical importance for some people. <eyalroz>ekaitz: I gave you a rocket emoji... :-) <eyalroz>I wonder - if there are known bugs in GCC, and some of them I suppose apply to GCC 4.6 or 4.7 and have only been fixed in later versions - does that not mean there is some chance of something being wrong with a newer GCC version compiled with one of them? <eyalroz>Also, if newer-GCC gradually starts using constructs from newer versions of the relevant language standards (C, C++) - doesn't it become unbuildable with GCC 4.6/4.7 ? Or is the idea to build a sequence of progressively-increasing-versions of GCC? <ekaitz>more modern ones need this intermediate step <ekaitz>about bugs, yes, they can happen, but life is hard <ekaitz>ACTION is not a compiler engineer <stikonas>even GCC 10 can be built with some light patching <stikonas>but we have ocasionally seen bugs that do propagate <stikonas>when ekaitz and I initially bootstrapped tcc on riscv <stikonas>it was buggy and no amount of recompiling would fix it <stikonas>that buggy version could build a lot of stuff <ekaitz>what a time! we enjoyed a lot didn't we? <stikonas>lots of utilities such as gzip, tar, make etc... <stikonas>that were not present if you build the same tcc source with gcc <ekaitz>some of them were triggered by the second compilation <ekaitz>first compiler is wrong -> next compiler produces weird code -> next program fails