[Development] Removal of xlib plugin - a possible disaster.

[lars.knoll at nokia.com]

None of the below changes the fact that the xlib plugin is unmaintained
and broken. Removing it is the right thing to do for exactly that reason.

If things don't work with an older xcb release, let's add a configure test
requiring a minimum version that works. I rather have configure bail out
with an error message telling the user to upgrade xcb (or their
distribution) than advising them to use a plugin we know is broken.

Cheers,
Lars

On 3/18/12 7:59 AM, "ext Rick Stockton"
<rickstockton at reno-computerhelp.com> wrote:

>http://codereview.qt-project.org/#change,20091
>My negative review of the Change comes out of 3 facts:
>
>(1) I have personally experienced the yet-to-be-documented problem, with
>both widget and declarative-based example programs, in which the GUI
>never appears: The program crashes with SIGSEGV, probably when
>attempting to assign size hints for the main Window. For me, the problem
>began occurring in February- but I wasn't doing much with Qt on XCB
>platform plugin during late January, so the problem may have been
>created/exposed at an earlier time.
>In any case, it was happening 100% of the time with my Distro, Mageia-1.
>(I have new problem with Mageia-2, described as "fact #3" below.)
>
>(2) That problem has been reported by others on IRC. I'm not capable of
>creating a "tight" search on Jira bugreports, but I didn't see a report
>for the problem. If someone can assist in that search, we'd be looking
>for xcb plugin, sigsegv crash, bug opened after the start of 2012.
>
>(3) Wondering if my Distro was "too old", (and most of it's RPMs were
>built in April of 2011), I just upgraded to Mageia-2 "beta-2". I no
>longer get SIGSEGV, but I get a problem which _could_ be closely
>related: referencing an Undefined Symbol related to WM hints.
>https://bugreports.qt-project.org/browse/QTBUG-24835
>
>--- comments ---
>I can't think of a WORSE spectacle, for the reputation of Qt-Project,
>than the Release of an Alpha Build in which large numbers of exerpienced
>Qt users on Linux Linux _might_ be unable to use the XCB plugin and show
>a main Window, before getting some kind of 'ABEND', with _any_ of our
>own provided examples. Sending the Alpha out with xlib plugin removed
>would be like selling a car without an emergency brake: Yes, the main
>brake system is the one you should use, but if totally dies .... you
>have GOT to have another one.
>
>With these two bugs present (i.e. SIGSEGV happening on some "older"
>Distros, and "Undefined Symbol" happening on RPMs built just 6 days
>ago), you NEED to have an alternative, EASY plugin for these people to
>use. Even at alpha.
>
>Yes, the bugs should be documented, isolated, and fixed. But, from my
>own Test/Integration/Release career, I'd say that our choices for alpha
>are (A) leave the xlib plugin present, and maybe even make it build
>automatically on X11 systems; or (B) stop ignoring the IRC "I'm getting
>'segment fault' with an example program" postings, and POSTPONE ALPHA
>until we've got them fixed. And I'd vote for alternative A -- keep the
>alpha schedule, but provide xlib as an alternative plugin.
>
>Just my opinion, but it's held pretty strongly. Since I have only ONE
>box, with ONE disk, I can't easily go back to mageia-1 to reproduced bug
>#1. But I'm pretty confident that some of our Qt5-Alpha-1 users will
>experience it, and their doco might help to pin it down pretty well. :)
>_______________________________________________
>Development mailing list
>Development at qt-project.org
>http://lists.qt-project.org/mailman/listinfo/development