Bug 3389 – gdb: using -gc is mandatory but not well documented

-g is known to be broken when using GDB (I don't know if this is how it's supposed to be or if it's a bug). -gc usually works (there are a few bugs still), but it's not very well documented that when using GDB you should use -gc (which is supposed to mean, "pretend to be C", so it's not very easy to figure out that that flag should be used instead of -g).

-g not working is how it's meant to be, at least until gdb adds support for the D extensions to DWARF. I've hopefully fixed the remaining bugs with -gc (on linux at least). My solution to this would be to add in a -gd, and make -g an alias to -gc until better support for debug info is added to debuggers. This way users get working debug output with -g, and don't blame it on a buggy dmd :)

Seems reasonable.

Copying comment in bug 4149 and reopening because I don't think this is really closed. In that bug some standard DWARF issues were moved to -g, and now I think -g works with debuggers not supporting D extensions. *BUT*, in bug 4149 Walter agreed about this comment by Brad Roberts: > Depends on if gc implies c, c++, or 'as much as is supported in the built-in > debug format without extension'. > > I'd argue for the last definition. Given that dwarf supports it without > extension, I'd argue that it should use it. I'd argue that -g should be > built-in + d extensions. AFAIK this pull request works for -g instead of -gc, even when it uses standard DWARF features, which goes against that comment. Also there is the point made by Robert Clipsham in comment 1, which for me makes a lot of sense. At least this will be more familiar with people used to GCC command line arguments (which is probably 100% of the *nix world), where -g is the default for debug and you have, for example -ggdb for GDB extensions. AFAIK there is no clear direction about this and it would be nice to have one, so the debugging stuff is as consistent as possible.

Thats a wrong comparision because neither -g not -ggdb make the compiler fake C++ debug information as C. The -gc flag is a hack to masquerade as C which should not be the default if -g works for most platforms.

GDB already has D specific extensions (demangling and arrays). Using -g makes use of them, -gc does not. There won't be D specific DWARF extensions any time soon. What we wanted to add can be modeled by newer DWARF versions but adapting them would break older debuggers.

The bottom line is that -g needs to 'just work' with the standard debuggers on the supported platforms. Assuming that -g works, -gc is a left over appendage. A whole lot of progress has been made in a number of commits over the last year. Some quick testing on linux shows that it works quite well with just -g now. With that in mind, I'm resolving this bug report as wontfix since -gc isn't mandatory any more. Chances are that there's more bugs left, so please feel encouraged to distill repro cases where -g doesn't "just work", file them, and we'll work on them.

(In reply to comment #4) > Thats a wrong comparision because neither -g not -ggdb make the compiler > fake C++ debug information as C. The -gc flag is a hack to masquerade as > C which should not be the default if -g works for most platforms. That's not what Walter said (agreeing with Brad's comment) in the comment I pointed out. But it seems that's not longer the position about -g/-gc (at least for Brad). I don't care that much about what should be the meaning of -g/-gc as long as there is one that works (and will keep working) with standard debuggers. (In reply to comment #6) > The bottom line is that -g needs to 'just work' with the standard debuggers on > the supported platforms. Assuming that -g works, -gc is a left over appendage. > A whole lot of progress has been made in a number of commits over the last > year. > > Some quick testing on linux shows that it works quite well with just -g now. > > With that in mind, I'm resolving this bug report as wontfix since -gc isn't > mandatory any more. > > Chances are that there's more bugs left, so please feel encouraged to distill > repro cases where -g doesn't "just work", file them, and we'll work on them. OK, so -gc should be just deprecated and there should be only -g, right? Will -gc be the same as -g? If not, why?