(In reply to Steve Fink [:sfink] [:s:] from comment #14) > Which are the gcc toolchain builds? I'm guessing they're using a different version of gcc. These are the builds shown as `Bbc` in a try build. They use gcc 7.4.0 at the moment. > I get the problem using the gcc 8.3.0 that the hazard analysis uses, just using a straight gcc compile without the hazard analysis included. That file compiles successfully with my system's gcc 9.2.1. > > It seems to be getting the wrong type for the `mTransaction` field. I guess it must be some bizarre gcc bug involving template instantiations (maybe) and `using Class::fieldName` declarations (probably). > > I'm checking to see what happens if I try to recompile the plugin for gcc 9 right now, but that'll probably involve additional effort. I will check if I can manage to get a gcc 8.3.0 to reproduce this and make some workaround for now. But in general it would be great to update the hazard analysis to a newer gcc version (when we move to C++20 at some point it will be inevitable).
Bug 1623278 Comment 15 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
(In reply to Steve Fink [:sfink] [:s:] from comment #14) > Which are the gcc toolchain builds? I'm guessing they're using a different version of gcc. These are the builds shown as `Bb` (not `Bbc`, these are builds using an older clang version) in a try build. They use gcc 7.4.0 at the moment. > I get the problem using the gcc 8.3.0 that the hazard analysis uses, just using a straight gcc compile without the hazard analysis included. That file compiles successfully with my system's gcc 9.2.1. > > It seems to be getting the wrong type for the `mTransaction` field. I guess it must be some bizarre gcc bug involving template instantiations (maybe) and `using Class::fieldName` declarations (probably). > > I'm checking to see what happens if I try to recompile the plugin for gcc 9 right now, but that'll probably involve additional effort. I will check if I can manage to get a gcc 8.3.0 to reproduce this and make some workaround for now. But in general it would be great to update the hazard analysis to a newer gcc version (when we move to C++20 at some point it will be inevitable).