This is the mail archive of the
mailing list for the GDB project.
Re: [patch+7.4] reread.exp 7.3->7.4 regression
On Mon, 19 Dec 2011 20:04:30 +0100, Tom Tromey wrote:
> Perhaps Pedro's problems with that patch could be alleviated by having
> "destroy in place" and "initialize in place" functions that both
> free_objfile and reread_symbols could use.
The problem is that if one uses file A with separate debug info and new file
A no longer has split debug info then one (or more on OSX) objfile will need
to be removed anyway. So `struct objfile *' cannot work.
Maybe one could make a compromise it is valid to hold `struct objfile *' iff
it is not a separate debug info file and review all the `struct objfile *'
holders with this rule.