On 26 April 2012 12:12, Dodji Seketeli <do...@seketeli.org> wrote: > Manuel López-Ibáñez <lopeziba...@gmail.com> a écrit: > >> On 21 April 2012 14:56, Jason Merrill <ja...@redhat.com> wrote: >>> It seems like we'll do this for every line in the header, which could lead >>> to a lot of leaked memory. Instead, we should canonicalize when setting >>> ORDINARY_MAP_FILE_NAME. >> >> Hum, my understanding of the code is that this is exactly what I >> implemented. That is, at most we leak every time >> ORDINARY_MAP_FILE_NAME is set to a system header file (if the realpath >> is shorter than the original path). This is a bit inefficient because >> this happens two times per #include (when entering and when leaving). >> But I honestly don't know how to avoid this. > > My understanding is that by the time we are about to deal with the > ORDINARY_MAP_FILE_NAME property of a given map, it's too late; we have > lost the "memory leak game" because that property just points to the > path carried by the instance _cpp_file::path that is current when the > map is built. So the lifetime of that property is tied to the lifetime > of the relevant instance of _cpp_file. > > So maybe it'd be better to canonicalize the _cpp_file::path when it's > first build? One drawback of that approach would be that > _cpp_file::path will then permanently loose the information about the > current directory, that is indirectly encoded into the way the file path > is originally built. But do we really need that information?
Another drawback I didn't realize until now is that in this way the canonicalize every path, instead of only touching those that belong to system headers. I am not sure if it is important or not. Comments? Cheers, Manuel.
