Denys Vlasenko <dvlasenk(a)redhat.com> writes:
On Mon, 2011-06-27 at 14:17 +0200, Jiri Moskovcak wrote:
> On 06/27/2011 01:19 PM, Denys Vlasenko wrote:
> > On Mon, 2011-06-27 at 13:11 +0200, Denys Vlasenko wrote:
> >> On Mon, 2011-06-27 at 11:05 +0200, Nikola Pajkovsky wrote:
> >>> This code belongs to abrt.git.
> >> Why do you think so?
> > I would like to expand a bit.
> > Having cli functions split into two binaries in two different packages
> > makes it a bit harder to work with it, both for developers
> > (two different source directories) and for users (they need
> > to remember which operation they can do with which tool
> > - report-cli or abrt-cli).
> > This split has to have some advantages which outweigh this.
> > What are they? Off-hand, I don't see any.
> Just from the top of my head:
> - abrt-cli needs to know about /var/spool/abrt which is "internal" abrt
> knowledge and any part of libreport should know about it
(I assume you mean "should not know about it").
report-cli currently uses /var/spool/abrt as a default for -D DIR.
If you feel that having it is "incorrect", then I have a few ideas
how to make it "more correct":
This default can be (1) removed, or better, (2) made configurable at
build time, say, configure --with-default-dump-dir=DIR.
For Fedora, we'll build it with --with-default-dump-dir=/var/spool/abrt.
Whoever else wants to build it, they can use anything they want,
or omit it and have no default.
How does this look to you?
Why don't make it configurable from abrt.conf?
I think a lot of people don't want to bother with changing spec file (if
you want to have exact version of abrt/libreport as is in rhel or fedora)
Save the whales. Feed the hungry. Free the mallocs