Parsing a log with the MSBuild parser takes over 10 minutes in severe cases, whereas the GCC parser takes at most 10 seconds on a similar log. I'll admit that the log is rather large (17 Mb), but the performance between GCC and MSBuild differs by a factor of 60.

This log is the result of a Pipeline build which makes 14 parallel sub-builds on different compilers and platforms, and one possibility to lessen the severity of this issue would be to be able to run the parser on just the current "node" or "stage" statement in the Pipeline script. This would prevent the MSBuild parser from having to see the gazillion of lines coming from GCC, which it won't recognize anyway.