>>> Gah. I was going to commit this separately but forgot. Sorry about that. >>> >> K, why does this fix the test? seems weird that we had an infinite >> loop before and changing the flags gets rid of it? > > It wasn't an infinite loop but an ICE in the tree caches. The > different options were producing different ASTs, so the tree cache had > different trees on write-out and read-in. >
Ah ok I see, this is actually the test I was aiming at fixing next, nice. I remember we were discussing it was the build system's job to ensure flags matched. In this case however it's much worse then the asm diff we were getting before in the c1builtin-integral test when using different flags, which was fine as the assembly was still valid and was just adapting to the behavior wanted by the user. An ICE in lto_streamer_cache_get is not very intuitive to point out the problem to the user in this case imo.... I'm thinking we should at least warn when some flags like this one differ and when that difference has an impact on the internal representation and can produce ICEs. Gab
