Re: drkonqi integration

Am Freitag, 26. Juli 2024, 17:40:02 MESZ schrieb Igor Kushnir:
> On 2024-07-25 12:11 PM, Harald Sitter wrote:
> > Yo
> >
> > there is this pile of code that lets kdevelop interface with drkonqi
> > to hand over the debugging from drkonqi into kdevelop. it currently
> > has no UI backing in 6 and also doesn't really work with the default
> > coredumpd-based crash handling
> >
> > I am wondering how sad we would be if we just threw away the code?
> >
> > At a glance we could remove
> >
> > - the entire socket tech in kcrash
> > https://invent.kde.org/frameworks/kcrash/-/blob/master/src/kcrash.cpp?=
ref_type=3Dheads#L799
> > - the external debugger tech in drkonqi
> > https://invent.kde.org/plasma/drkonqi/-/blob/master/src/debuggerlaunch=
ers.h?ref_type=3Dheads
> > - dbus interaction code in kdevelop
> > https://invent.kde.org/kdevelop/kdevelop/-/blob/master/plugins/debugge=
rcommon/midebuggerplugin.cpp#L135
> >
> > HS
>
> Hi everyone!
>
> Adding the author of the most recent work on this integration:
> https://invent.kde.org/kdevelop/kdevelop/-/merge_requests/202
>
> Christoph, would you like to fix the DrKonqi integration again, now or i=
n the
> future? If you don't have motivation or time for this, before removing i=
t all,
> we need to consider how difficult fixing the integration will be and how=
 often
> it is used. As the integration only broke recently, perhaps wait for bug=
 reports
> about it against KDevelop 6.0?
>
> Cheers,
> Igor

Hi,

Actually the button that featured launching the debugger does not show up,
since QML is put in place for DrKonqi. That's the case for some time now.
Should be effective since commit 61e4560e (grand qml UI rewrite) from '22.
If anybody missed the button, there should have been bug reports already.

I liked investigating crashes quickly when they appeared, but lost the mot=
ivation
to do so since the button was gone. I think only a handful of people (mayb=
e just
me) used the feature however. So cleaning that up would probably be the be=
tter
choice to reduce complexity. Especially considering the effort it takes to=
 revive
the feature (together with the opt-in gdb debug possibility that nobody kn=
ows
about).

Rephrasing Haralds question: Would anybody in this mailing list like to se=
e the
button again?

Best regards
Christoph