Opened 5 years ago
Closed 5 years ago
#12631 closed defect (fixed)
Fix mutter-3.34.0{,1} on i686 systems and update to 3.34.3
Reported by: | Douglas R. Reno | Owned by: | Douglas R. Reno |
---|---|---|---|
Priority: | normal | Milestone: | 9.1 |
Component: | BOOK | Version: | SVN |
Severity: | normal | Keywords: | |
Cc: |
Description
From tnut in ticket #12617:
On a 32 bits system I had to pass: export CFLAGS="-Wno-pointer-to-int-cast" to make mutter 3.34.0 compiling successfully.
Change History (8)
comment:1 by , 5 years ago
Owner: | changed from | to
---|---|
Status: | new → assigned |
comment:2 by , 5 years ago
comment:3 by , 5 years ago
Still have the same issue with meson-0.52.1. I've not found any workaround, except downgrading meson to 0.51.2. Using ninja -d explain, I get:
ninja explain: output build.ninja older than most recent input ../../../../dev/stdout (1575663778421193578 vs 1575663778501192571)
Which is nonsense, since there is no reason for imposing stdout to be older than build.ninja (and actually that cannot be true)
comment:4 by , 5 years ago
First as short as I can: I've been able to solve the issue by adding -Dcogl_tests=false.
Some explanation: the only places in the source tree where /dev/stdout is referenced are two files meson.build in subdirectories of cogl/tests. Specifically, '/dev/stdout' appears as an argument of a run_command() meson function. I guess that when a file appears in such a command it is automatically added as a dependency... But having /dev/stdout as a dependency when redirecting output is something silly. Well, just a guess. Could be something else.
comment:5 by , 5 years ago
Summary: | Fix mutter-3.34.0{,1} on i686 systems → Fix mutter-3.34.0{,1} on i686 systems and update to 3.34.2 |
---|
Merge with #12968
Almost there
comment:6 by , 5 years ago
Summary: | Fix mutter-3.34.0{,1} on i686 systems and update to 3.34.2 → Fix mutter-3.34.0{,1} on i686 systems and update to 3.34.3 |
---|
Now 3.34.3
comment:7 by , 5 years ago
mutter-3.34.2 change notes:
3.34.2 ====== * Fix immediate screen blank after releasing inhibitor [Tim; #573] * Respond to frame callbacks regardless of damage [Jonas Å.; !839] * Translate well-known selection atoms to mimetypes [Carlos; !842] * Fix Night Light on wayland [Jonas Å.; !840] * Fix various copy+paste/DND regressions [Carlos; !848, #789, #842, #793, #845, #854] * Don't emit focus event after desctruction [Marco; gnome-shell#1704, !860] * Fix hang when opening not-responding dialog on Xorg [Carlos; !876] * Fix frozen grabs on Xorg after weeks of inactivity [Jonas Å.; !886] * Fix triggering popups from stylus devices on wayland [Carlos; #886] * Support shadow framebuffers for offscreen rendering [Olivier; !917] * Fix hang after interacting with desktop icons on X11 [Marco; !909] * Don't double scale when getting absolute surface coordinates [Xiang; !915] * Kill window effects on destroy [Robert; !924] * Use partial damage for dma-buf and EGLImage buffers on wayland [Robert; #947] * Do not stack transients underneath their always-on-top parent [Florian; #587] * Fix KMS freeze after pageflip [Pekka; !953] * Fixed crashes [Robert, Jonas Å., Marco, Hans, Carlos, Tim; !856, !912, !895, !928, #591, !823, !960] * Plugged memory leaks [Niels, Robert, Carlos; !847, !868, !873] * Misc. bug fixes and cleanups [Daniel, Marco, Jonas Å., Georges, Cosimo, Florian, Hans, Robert, Jonas D.; !841, !764, !837, !846, !673, !811, !893, !925, !927, !940, !832] Contributors: Marco Trevisan (Treviño), Cosimo Cecchi, Tim Crawford, Jonas Dreßler, Xiang Fan, Olivier Fourdan, Carlos Garnacho, Hans de Goede, Niels De Graef, Tim Klocke, Robert Mader, Florian Müllner, Georges Basile Stavracas Neto, Pekka Paalanen, Daniel van Vugt, Jonas Ådahl Translators: Bruce Cowan [en_GB]
mutter-3.34.3 change notes:
3.34.3 ====== * Fix window recording on HiDPI [Pascal; !976] * Fix top-left pixel being insensitive to clicks [Sebastian; #893] Contributors: Sebastian Keller, Pascal Nowack
While we are at it, I think we should check this package with meson-0.52 (it builds fine with meson-0.51.2). When running ninja , I get:
I do not know exactly where it comes from. I think the problem is that, in build.ninja, the build target build build.ninja: depends on meson-private/coredata.dat, which is generated at the same time build.ninja is generated. So if by chance the modification time of meson-private/coredata.dat is posterior to that of build.ninja, it is regenerated.
But I have tried to remove meson-private/coredata.dat from the list of dependencies (with a sed before running ninja), and it goes a little farther, that is, it compiles everything before starting again:
Note that it may be a race condition. All the above occurs when scripting and redirecting output to a file. If I manually run the commands and have the output go to the screen, the build file is regenerated only once by ninja, then compilation starts...