[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Release schedule
From: |
Alexander Malmberg |
Subject: |
Re: Release schedule |
Date: |
Wed, 02 Apr 2003 00:56:32 +0200 |
Willem Rein Oudshoorn wrote:
> Alexander Malmberg <address@hidden> writes:
> > No, but it depends on what kind of crash reporter is intended.
>
> Well, this brings me to the following idea.
> We have code in base that gives the returns the return addresses
> of the current stack frame.
Even if it can grab the immediate return address, it's a much more
difficult problem to walk backwards and get the other return addresses
that are part of the stack trace. I'm not sure how capable this code is,
though.
> gdb and the addr2line utility can translate these addresses to
> filenames and linenumbers.
Only if the programs and libraries were compiled with debugging
information, which is not the case by default. (I've hacked my local
copy of -make to always compile with -g, but there is a significant disk
space hit associated with this.) You can still usually get the function
names, which is better than nothing.
> If we can combine these two inside base we could put
> as extra information in the exceptions the stacktrace
> with filenames and line numbers.
>
> This will help with debugging.
Automagically invoking gdb or something to grab filenames and line
numbers would be a big help in many cases. (Not as much as a clear,
"working" description of how to reproduce a problem, though. :)
- Alexander Malmberg
Re: Release schedule, Tim Harrison, 2003/04/01
Re: Release schedule, Yen-Ju Chen, 2003/04/01
Re: Release schedule, Willem Rein Oudshoorn, 2003/04/01
Re: Release schedule, Adam Fedor, 2003/04/02
- Re: Release schedule, David Ayers, 2003/04/02
- Re: Release schedule, Alexander Malmberg, 2003/04/02
- Re: Release schedule, Markus Hitter, 2003/04/03
- Re: Release schedule, Chris B. Vetter, 2003/04/03
- Re: Release schedule, Richard Frith-Macdonald, 2003/04/04
- Re: Release schedule, Chris B. Vetter, 2003/04/04