This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [RFA] diagnostic.c: Emphasise preprocessed source for bug reports


On Wed, 24 Jan 2001, Neil Booth wrote:

> We get a large number of bug reports without preprocessed source.
> There is little point asking people, via "feedback", for preprocessed
> source more than 24 hours after the event -- I don't think anyone has
> ever responded (quite reasonably IMO - they've long since moved on) to
> such a request.  This renders many reports useless, that might
> otherwise be useful.

How about making the compiler produce the preprocessed file itself?
Integrated CPP makes this more complicated, but the compiler ought to be
able to communicate the fact of the ICE back to the driver, which could
then run the standalone preprocessor to create the preprocessed file (with
appropriate checks against looping if the ICE is in the preprocessor, and
a sanity check that the preprocessed file does reproduce the problem),
then tell the user the name of the preprocessed file to send (or, if it
can't produce a preprocessed file or the preprocessed file doesn't
reproduce the problem, a list of the names of all relevant source files).
Perhaps put this on your list of things to do in your specs rewrite - the
compiler has the relevant information and should be able to package it
itself rather than requiring the user to rerun a compilation command.  In
principle the compiler could also provide the option to cut the
preprocessed file after the point the parser had reached when the ICE was
detected, and even (with appropriate preprocessed-file-reducing options
rather than as part of automatic preprocessed file production) remove from
the file functions and declarations not needed to reproduce the ICE -
since it knows the division of the file into declarations and definitions
and which were used where.

-- 
Joseph S. Myers
jsm28@cam.ac.uk


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]