*** This bug has been marked as a duplicate of 79184 ***

re-opening on foot of Owen's comments in #79184. 

"Note that most of the dups on this bug are inappropriate. 
               The bug is about a Sun keyboard problem that occurs
wiht
               any app.

               There is a separate problem with gcalc where the keypad
               work on any machine. It could be a GTK+ bug"

we really should get this working for 2.0.1 - i need time to figure
out what it's doing

*** Bug 88494 has been marked as a duplicate of this bug. ***

Probably one of those 'should just work' things....

Was this one fixed? It seems to be working on Laca's "Sun" build from
22nd July (first build with long package names etc. to meet ARC reqs)

mmmm, this appears to be still broken in a build from the head on 25th
july

This is wierd - I have 2 builds that are identical except one is a
straight nightly and one is a sun stable (branded) nightly (both dated
30th July). 

In the nightly build this bug is still present - the "del" key on
numpad will delete any numbers added to gnome-calculator (with num
lock on or off). In the sun branded build with numlock on the "del" ky
will add "." (as expected) - i.e. this bug is not present.

AFAIK the only differences between the two builds are the location
gnome is installed to and the naming convention used for the package
names.

Glynn/Laca: Can you think of any reason why this would be happening. 

Shane, the "branded" nightlies have all the sun-patches (as in
CVS module sun-patches) and branding patches applied.
Note that there are several gtk+ patches in sun-patches.

I'm going to close this as fixed

verified

closing

Whoa there. Shane, am I understanding correctly that this works
correctly only in the Sun version with Sun patches? Then shouldn't we
reopen this bug until we've got those patches migrated back upstream
into GNOME cvs? Or do I misunderstand where the patch is that fixes this?

John: sun's chosen to take a Definitely Not Approved stance wrt gtk
and keyboards and keypress events (it's a long story; I'm sure someone
cc'd here can give more details on it) so this stuff is very much
sun-specific and not general. Sorry that's not more clear in this
specific report.

I don't think that's the right approach, regardless of Sun's internal
approach. The bug's still there, we've gotten a bunch of reports from
Sun and others, it's not fixed. I don't mind if Sun people consider it
fixed for their internal purposes if it's fixed with their internal
patches, but we shouldn't close it in GNOME bugzilla until it's fixed
in GNOME cvs, right? Or am I misunderstanding something? I'm moving
this to NEEDINFO for now so this doesn't get lost.

By 'not approved' stance wrt keyboards and gtk, I mean 'owen won't
take those patches, ever.' So the gtk fixes can't be put upstream. 

This particular bug (Owen claims it is not a duplicate, but I think he
was confused) works just fine on Linux, and works on Solaris with the
Sun gtk patches, so if it ever was a bug in gcalc it is now fixed.

Can you close if that's sufficient info, John?

The reason why this is working is that we have a local patch against
gtk+ in sun-patches that we are applying to our build. Now there is
definitely an intention to get this merge back into gtk+ as soon as
owen can review it...but from what I could gather from past irc
conversations it's also broken for several other people who definitely
aren't on Solaris machines....I'm not quite sure what to make of it
all - whether its a gnome-calculator problem, a gtk+ problem etc..
without the time to track it down [which I currently don't have] :(

Glynn: it currently works here on Linux, and Shane claims it works
with a patched gtk on Solaris, so... I'm really doubting it was a
gcalc problem, or if it was, it's been fixed at some point.

I'd rather see this one reopened and kept around until we sort this
out. This is one of a family of bugs related to numlock behavior in
gcalc (#85356 and #88880 being the siblings), with great confusion
over where and how they occur - what's in Solaris, what shows up
consistently across platforms, whether they're gtk or in the
calculator itself, etc.

P.S. It works for me (that is to say, there is no bug) on Linux with
the latest cvs build. :-)

John: this bug /is/ sorted out. It's not a bug anywhere anymore
(assuming one is running a 'fixed' gtk on Solaris.) So I'm going to
close it. The others definitely need to be open. 

So if what you're saying is that all Solaris users will be getting
Sun's fixed version, then I'll buy the argument that it's fixed. Is
that the case? Or will there be Solaris users who get plain GNOME
tarballs? Because the bug isn't fixed for them. (Forgive the ignorance
- I've no idea how the Solaris community gets its software. Maybe
that's the source of the confusion here. I've been operating under the
assumption that GNOME-distributed tarballs should be made to work for
Solaris users, but if all Solaris users get their stuff from the Sun
fixed version, then I've been a dolt here all along.)

/If/ the solaris user applies the /gtk/ patch, then they won't see a
problem. So keeping a gcalc bug open has no value, since gcalc does
not need to be patched or changed in any way.

So can we then keep a gtk bug open on this?

I hope you can understand my discomfort, Luis, with saying a bug has
been fixed when the fix lives somewhere else besides GNOME cvs. If
there's already a gtk+ bug discussing all this and codifying the patch
then again, I'm being a dolt.

I /believe/ 79184 is that bug. Not sure, though- maybe one of the sun
people could confirm?

I've asked Laca about this and all the sun/solaris specific patches
live in cvs in a dir called sun-patches (i think). I think it is
reasonable that solaris users who want to build gnome from cvs should
also download and apply any of the sun specific patches from
sun-patches in cvs. Obviously this would have to be documented (dunno
where would be best place).


p.s. Would there be a risk that adding sun/solaris specific patches to
cvs head might have a detrimental effect for non-solaris users?

OK, I'm fine with this then. I didn't understand that Sun patches live
in GNOME cvs. Sorry to be such a pest.

nay bother - all the best are pests... hehe:)

Guys: I'm re-opening this ...

Luis: you're making assumptions here and you shouldn't. Owen said that
this was a different bug from #79184 and it *is*. There are two
seperate patches for #85775 and #79184, so they must be seperate
problems. Secondly you're assuming that its okay for Solaris users to
have to apply these patches before building gtk+. That is also
incorrect .... And last wtf gave you impression that "owen won't take
those patches, ever." AFAIK, he's never seen this patch, and he only
has rejected other patches because he didn't see them as being correct.



Shivram: this bug needs to be moved to gtk+ and you need to attach
your patch here and explain to Owen what it does.

Right okay. I've just seen this in the status.xml file

=========
Fix for Keypad "." and Keypad Delete not working on Sun Type-4
keyboards The problem is due to the following and a bug against the
Sun XServer can be filed:

For Sun Type-4 keyboards a KP_Delete keysym is missing. The xmodmap
output for the keycode where one would expect a KP_Delete shows Delete
in non XKb mode.

          keycode  57 = Delete Delete KP_Decimal

When the XKEYBOARD extension is enabled the xmodmap output has a
KP_Delete keysym

          keycode  57 = KP_Delete KP_Delete KP_Decimal

Although in general, programs do not handle a Delete and KP_Delete   
     differently, the gnome-calculator does. The Keypad Delete and '.'
work as expected with keyboards where the keysyms for the
corresponding keycode are as expected.

The patch would work reasonably well and would have to be maintained
as a sun_patch till the time this issue has been resolved in the Sun
XServer.
=============

Shivram: this kind of information needs to go into bugzilla. Closing
this as NOTGNOME