libera/#clasp - IRC Chatlog
Search
13:07:33
drmeister
Bike: Hey - good morning. I have a zoom call this morning demoing the sequence analysis to our collaborators down in Florida.
13:18:40
Bike
<?https://github.com/gcc-mirror/gcc/blob/master/libstdc++-v3/libsupc++/eh_personality.cc#L497-L519?> here's the part where it parses the table, <?https://github.com/gcc-mirror/gcc/blob/master/libstdc++-v3/libsupc++/eh_personality.cc#L544-L643?> here's where it decides whether it matches, <?https://github.com/gcc-mirror/gcc/blob/master/libstdc++-v3/libsupc++/eh_personality.cc#L649-L662?> here's the result of a
13:42:59
Bike
this here is actually two completely different functions with different names and operations, depending on what is defined and what isn't
13:48:04
beach
I usually think it was pretty gutsy by RMS to choose C and Unix. I would have done the wrong thing by choosing Lisp and Genera, but then free software would not have existed. However, writing a compiler in C is just nuts.
15:21:50
drmeister
Bike: I have two sessions of clasp running under udb and one has gnu libunwind with debug symbols and the other has gcc_s with debug symbols.
15:22:33
drmeister
I can see the point where they diverge. But I have no insight (yet) into what is wrong.
15:24:12
drmeister
with gcc_s the lsda is simply returned. With libunwind it parses stuff and then returns an lsda - it's probably doing something wrong here.
15:24:35
drmeister
Comparing the lsda data is hard because I don't know what is address specific and what is invariant.
15:25:05
Bike
https://itanium-cxx-abi.github.io/cxx-abi/exceptions.pdf this has the LSDA format, if that helps
15:28:19
Bike
_Unwind_Context is an opaque type so I guess it can be different in different implemtnations
15:29:01
drmeister
So gcc_s gives me a working lsda and gnu libunwind calculates one. They are different - but are the differences significant?