Re: Pre-approval for scim NMU
On Sat, Oct 18, 2008 at 01:17:27PM +0900, Osamu Aoki wrote:
> Hi,
>
> I changed my mind. Please think again before doing NMU. I do not think
> this patch is needed.
>
> On Sat, Oct 18, 2008 at 11:29:26AM +0800, Deng Xiyue wrote:
> > On Sat, Oct 18, 2008 at 12:06:37PM +0900, Osamu Aoki wrote:
> > > Hi,
> > >
> > > On Sat, Oct 18, 2008 at 10:24:53AM +0800, Deng Xiyue wrote:
> > > > Hi Release Managers:
> ...
> > > > I'd like to prepare an NMU for scim to fix bug#434180[1], which causes lock
> > > If this patch is tested to be working fine and Release Managers OK this
> > > action, I am happy for you to upload this.
> > >
> > > Please also make sure not to cause negative impact to other dependent
> > > packages. (You have option to NMU or binNMU those too.)
> > >
> >
> > The patch is specific to scim and shouldn't break reverse dependent
> > packages, at least not the ones I use (e.g. scim-pinyin).
>
> Does this patch used in upstream VCS or distribued in other distro
> (ubuntu? redhat?). How is the BTS on their sites?
>
> It seems to discard keyboad queue. I have no idea what impacts it
> causes. It is high risk unless it is tested widly.
>
> Unless you have facts to support goodness of this patch on several
> popular platforms of SCIM (not just one chinese IM), how can we be sure
> this does not break things. I was expecting a bit more testing etc.
> from you.
>
Indeed. The really problem resides in libX11, and the libX11 patch in
the same bug report is the real fix. The one I mentioned is just a
workaround in scim side. See below.
> As I reread your bug report, massive tab is the case you expeienced
> input locking. Is this SCIM problem? Are you sure? Did you make some
> analysys using GDB to prove it?
>
> I have sort-of freeze situation sometimes but that is when I have
> keyboard accessibility enabler capability via keyboard option set. Of
> course by discarding input queue, system will not cause such problem but
> that is not right thing to do. Just unclick gnome option for that or
> carefully shrink windows to find prompt from accessibility program and
> disable its activation. (I know it was annoying ...)
>
> As I also see that libX11 issue. It has been updated since then. Do
> you know the situation?
>
Sorry for this, as I'm not subscribed to bug#437437, which I should, and
#434180 is not blocked by it, I wasn't aware the proper fix has gone in
libx11, where the really cause located. That should be sufficient for
this problem. As since then, I switched to scim-immodule and
scim-bridge, I haven't tried xim yet. Thanks for this information.
According to my test in the last hour using unpatched scim, those common
approach to reproduce the problem are invalid now. Plus the feedback in
launchpad bug report, and other BTS such as fedora, I think now this bug
can be reassigned to libx11 and marked as done with
Source-Version: 2:1.1.4-1
As for the accessibility problem you mention, it might because the fix
for libx11 side is incomplete, but I'm not the expert in this regard.
> I have to say that I may not be the most technical maintainer of this
> package thus very conservative to deviate from the upstream. The
> information provided is not sufficient to make last moment upload.
>
> I am open to change my mind when sufficien support facts are presented.
>
I appreciate your precaution, and hopefully this bug has already been
fixed. So please consider this mail as withdrawal of my scim NMU
pre-approve request. Thanks Osamu.
> Osamu
Reply to: