State of the 1.9 release

[Peter Hutterer]

On Wed, May 19, 2010 at 11:42:50PM -0700, Keith Packard wrote:
> 
> We're closing in on the scheduled date for the merge window closure and
> I thought it would be a good idea to update people on what's going on.
> 
> Here's some patches that I know are outstanding:
> 
>  * mi: removed the invisible cursor sprite; use NullCursor instead.
> 
>    Oliver was trying to check with Coverity to verify if this works.
>    I'm frankly tempted to just merge it and have people test it; it
>    seems like a desirable API change in any case, so any bugs it
>    introduces should be fixed.
> 
>  * Turn deprecated allocation functions into inline ones
> 
>    This requires a new libXdmcp; I think Mikail is off on a trip soon;
>    so this may wait until he returns or someone else does the libXdmcp
>    release.
> 
>  * dix: Remove a large pile of backing store leftovers
> 
>    Just needs some minor fixups at this point, nothing major. Would be
>    nice if someone could review the autoconf patch that tries to
>    make miInitializeBackingStore appear for existing drivers. Otherwise,
>    meh.
> 
>  * dix: Reshuffle ScreenRec to pack holes
> 
>    Adam wants to play with this a bit more; it's not a big deal in any
>    case.
> 
>  * glxdriswrast: Use CreateGC, not CreateScratchGC.
> 
>    Bikesheding over the interface for creating long-term internal GCs
>    in the server. Others are welcome to join in the fray if interested.
> 
>  * Reimplementation of devPrivates
> 
>    The first of these two patches is huge and was done manually,
>    so there may well be errors. I'd love to see review, but failing
>    that, I'm very tempted to just get a few Tested-by: lines and push
>    expecting that we'll catch any latent bugs with the asserts() that
>    litter the new code.
> 
> If anyone has patches they think should be on this list, let me know.

I've got two input ABI breakers nearly ready. One adds the InputAttributes
to the InputInfoRec and the IDevRec so drivers have access to it. 
The other one adds a "flags" field to DeleteInputDeviceRequest so that
drivers know why a device is removed.

Both are quite necessary, we have some issues for drivers that implement
their own hotplugging scheme. Currently that's only wacom, but soon evdev
will be one of them. I hope to get both patches out tomorrow but Dan has
some InputAttributes patches as well that should go in around the same time
(to avoid bisecting nightmares).

> We've currently got 11 bugs on the release blocker:
> 
>  Bug 23938 - keys occasionally get stuck with xorg-server 1.6.99.901
>  Bug 23985 - Excessive core motion events for subpixel motions
>  Bug 24094 - CTRL-ALT-F1 doesn't switch to VT1 (provides garbage input to terminal instead)
>  Bug 24856 - xorg-server-1.7.1 mouse cursor not moving from MouseKeys
>  Bug 27804 - Enter/Leave event woes with multiple master devices.
> 
>      These are all input bugs. Surely Peter will have them fixed by 1.9.

yeah, right...

23985 is not a blocker, it's just annoying. 24094 is tricky, there's some
xkm stuff going wrong (which may also be related to what svu is seeing if he
exceeds 6 groups)

The others, well.. yes.

>  Bug 27232 - A small java program that causes X to hang
> 
>      Sure looks like it's just grabbing the pointer or keyboard;
>      probably not a bug.

yeah, I'm sure that's just unfortunate breakpoint placement but needs more
time than I can afford right now. If someone is willing to guide Mike
through how to use xscope and read protocol output, that' be much
appreciated.

> Secondly, it doesn't seem reasonable to close the merge window the day I
> return, so I think we should hold it open until June 4th instead, giving
> us most of a week to finish reviewing and stirring in new code before we
> get to RC1. That would change the schedule to:
> 
> Merge window closes:        2010-6-4
> Non-critical bug deadline:  2010-8-1
> Release:                    2010-8-20

can you please move those dates in the calendar too? I think I gave you
admin access when I set it up.

Thanks for this update, much appreciated.

Cheers,
  Peter