Re: Flex/Bison standardized diagnostics

[Akim Demaille]

>>>>> "Hans" == Hans Aberg <address@hidden> writes:

>> The GNU Coding Standards, section `Errors'.

Hans> Do you have a URL for this?

www.gnu.org

Hans> But there appears to be contradiction in your posts, because on
Hans> the one hand, you here just want just one error reporting
Hans> standard to be followed, and in the other post "Locations
Hans> Suggest" you say it can't be done, because there so many
Hans> variations. So how do want to have it?

Bison and Flex do not have to push the user towards a specific model,
but _they_ have to follow it.

Hans> I merely listed the things that might be needed in various
Hans> applications, and clearly stated that Flex/Bison may or may not
Hans> support that.

What matters is that it is possible, not that they are tuned for
CodeWarrior.

Hans> Some of the information is helpful when developing for IDE's,
Hans> and RMS said he wanted GNU to develop one such (as one of the
Hans> main reason to not use GCC is the absence of such). IDE's exist
Hans> on all the platforms UNIX, MSOS, MacOS, and may also be a part
Hans> of GNU.

We have this IDE: GNU Emacs.  That's why the error messages have this
form.

I am already working for an extension of this standard to make it more
accurate.  But that's a long walk: 1. have Bison be improved to
support *any* location scheme, 2. explain it to RMS, 3. have Emacs
understand it, 4. have GCC and others adjusted.

That's a long walk, which has started long ago.  Stop pushing us,
please.