| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
| |
It was fixed following v5.6.1, but was not merged into 5.7.0, only 5.7.1
will contain the fix.
Change-Id: Iaeb300a1b2eb93c4984ed39439903773736201f3
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|
|
|
|
|
|
|
|
| |
If <sys/user.h> is included before other system includes, compile errors for missing types
appear on FreeBSD 9 and 11. Tested compiling on FreeBSD 9 and 11 with 1.5.2 packages in
poudriere build jails.
Change-Id: I3669a14c01fb8d11f14e103c008f6f9cb33f7307
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
[ChangeLog] The "osx" value in qbs.hostOS and qbs.targetOS has been
replaced with "macos". For backwards compatibility, qbs.hostOS will
still contain "osx" in addition to "macos", and an error will be printed
if qbs.targetOS contains "osx" but not "macos". Specifying "macos"
without "osx" is allowed. These checks are subject to be removed in a
future version of qbs, so update your hostOS and targetOS checks
accordingly.
[ChangeLog] cpp.minimumOsxVersion has been deprecated and replaced with
cpp.minimumMacosVersion. cpp.minimumOsxVersion is subject to be removed
in a future version of qbs, so update your projects accordingly.
Change-Id: I479891829dff6eb6750cb2a04e1395f085896f63
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
|
|
|
|
|
|
|
| |
Task-number: QBS-983
Change-Id: Id0bd4a91f6afbe04621cfff28b0ef43dcfeee69b
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
| |
Otherwise, we would potentially try to install artifacts that haven't
been built yet, causing invalid error messages.
Change-Id: I5e8443e477ae767aeb6846de29dd8a3bd98b9ff5
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
| |
Task-number: QBS-984
Change-Id: I4324e387ec4124f3f5f638ab8d627fc33f6f21a9
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
| |
Task-number: QBS-985
Change-Id: If25310b37ee35cf786ae8d0c19d2196129ac2ab1
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
|
|
| |
Backport of 3432742caa5171ceddee0e75adefda9310c3b1eb in
qtbase/src/corelib/io/qlockfile_unix.cpp of QLockFilePrivate::processNameByPid()
that merged NetBSD, OpenBSD and FreeBSD as well as GNU/kFreeBSD
implementations into one block.
Change-Id: I5fdc086d5c2680d5a431339873c8685b0e68bf57
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
| |
Change-Id: I3a2361865036573237f1f87e2967f149bbf218f2
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
|
|
|
|
|
|
|
|
|
| |
That is, in the top-level file. Without this patch, such search paths
would not make imports available in the first file that was read, which
is not what users expect.
Change-Id: Iab62e046450a4677c8b7b68a1ad5ba70502826bc
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
| |
Change-Id: I93af50141a6180fb6663f5ee5d1463f57a56bd9e
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Abovementioned patch is completely broken. Now e.g. version
validators with min == max == 2 always cause errors for correctly
formed version numbers.
No more patches in this area will be accepted without an autotest.
I also strongly suggest not to throw hard errors anymore for this
kind of thing.
This reverts commit a9f70a366fdc47d58057116ae70c2d20dfa9c5f3.
Change-Id: I4d9c803cc2f914f05b888ea90c13a686479c4f96
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
|
|
|
|
|
|
|
|
| |
It seems that the {0,} specifier is equivalent to {1,}, so work around
this.
Change-Id: I06b6b6a7eb9561f948c0b0ba6e83a66800234e74
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If we
- held an existing project open (in an IDE) and
- changed its build directory and
- a second qbs instance also held a project open there and
- that second qbs instance used the same profile
then we would correctly detect that a competing build graph was already
present, but afterwards we would erroneously delete the build graph lock
object of the existing project, leading to a double-delete later when
the project itself was deleted.
Task-number: QTCREATORBUG-16376
Change-Id: Ie6c621f1dab5cc7b7ff97bf6c25d62609dc9eb35
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
| |
Change-Id: I885d94bb14b325dc36767a840ebdb0be1fb59dd2
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
|
| |
This term is not used by anyone (it's always written as "Qbs" and
pronounced "cubes") and it is ultimately confusing to users and
detracts from the identity of the product, especially its pronunciation.
Change-Id: Ifb76d059cafc33ce3fa686ddb3d7d67f47084faf
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We asserted on the condition that a node's product must be in the list
of products to build, meaning that it needs to belong to the set of
user-selected products or one of their dependencies. However, there are
at least two ways in which one can trigger this condition to be false:
1) The user provides a faulty project in which a "Depends" item is
missing, but the respective artifact from the other product is still
found (e.g. a generated header file via a project-global include
path).
2) The project is actually okay, but our C++ scanner erroneously adds
dependencies from a different product, e.g. because it does not know
about #ifdefs.
Instead of the assertion, we now simply mark the respective node as
built and continue. Ideally, we'd log a warning, but we cannot do that
because of point 2) above.
Change-Id: I3549d732dea5cde84d1019132580a8e051c9db11
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
| |
Task-number: QBS-980
Change-Id: I87f7cccf5a0e99f4c1762c56680f66c1a1b058b3
Reviewed-by: Orgad Shaneh <orgads@gmail.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
| |
Task-number: QTCREATORBUG-16346
Change-Id: I2ab57a0da5cfc2b2fc59898a80d2122b56b96685
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
| |
Change-Id: I7172c5afb0fb831b733d352c328fe57fdd75c179
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
|
|
|
|
|
|
|
|
|
| |
Incidentally, this also fixes a bug causing Windows 10 to be treated as
Windows 1.0 in the subsystem version.
Task-number: QBS-978
Change-Id: I5aa1763ce9d0a9b12aa50e864086cc2ae4ceb3c1
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
| |
Task-number: QTBUG-53392
Change-Id: I6691c9d6d3bb23f84b0739cd4a8f505be4c25cd2
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
|
| |
We overlooked them in a50456dcb0.
Change-Id: Iaa4a0094727209c30f385521d48fb70e43a2bd25
Reviewed-by: Eike Ziller <eike.ziller@qt.io>
|
|
|
|
|
|
|
|
|
|
| |
If the probe runs successfully (i.e. "found" is true), all output
properties should have defined values so that users can work on them
directly rather than first having to convert "undefined" to "empty
array" etc.
Change-Id: I96a380480ea746cc0413553e49097f1e417b71af
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If a product in a project with an extra search path exports a dependency
to a module that is found using that search path, *and* a second product
that is located in a project without the extra search path depends on
the first product, *and* a third product that is located in the project
with the extra search path also depends on the first product, we
inexplicably ignored the failure to load the exported module, leading to
an assertion later on.
Change-Id: Id0e4a67e912ded679fa35ec397b51dcaabd65cce
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
| |
Should not omit this in dealing with dynamic memory.
Change-Id: Ifc1e70a353446ab61023b9224a23285febfa6b97
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
|
| |
This does not seem to be reproducible currently (found through static
analysis), but it's fragile if the code structure of the callee changes.
Change-Id: I3500a631306f00392e83dbc83103b41e0bf2dc9c
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
| |
Change-Id: Icc4d9f5239ce1dc8fb8dd72bad4bdcfae1f51737
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
| |
Task-number: QBS-969
Change-Id: I0a2062415da31da02dd5e2013b377e90e8108278
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
|
| |
Do not hard-code the shell-quoting.
Change-Id: I4e7fafa28e0248581550a6467ce62f1c221d73c0
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
We must actually call some function from the setupapi lib to provoke a
link failure in case our static lib handling is broken.
Change-Id: I56444ca5f31e653c3f56b7fa33c1485ced797593
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We used to call certain member functions on an object that could be
null. While that was never allowed, it did not cause problems so far
because these functions checked their "this" pointer and returned
immediately if it was null. With GCC 6, such constructs no longer work.
Measures taken:
- Fix the null pointer accesses by allocating the respective
objects earlier (in case they must not be null) or checking for a
null value at the calling site (in case they can legitimately be
null).
- Rewrite the aforementioned member functions so that they will
no longer silently tolerate undefined behavior with other
compilers than GCC 6.
Task-number: QBS-974
Change-Id: If07ddc5c0e536d36c19311357f871554aa692ad6
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Reviewed-by: Ilya Lyubimov <villytiger@gmail.com>
|
|
|
|
|
| |
Change-Id: Ib859cc926a9f26dae1dda88b9f2cb23cac4439ee
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This did not happen before, because module conditions are only evaluated
when loading the module file, of which there is none in the case of
Export items.
Note: The implementation assumes that the condition of an Export item
has the semantics "use or ignore this Export item". The other possible
meaning would be to use the condition as the condition of the resulting
module instance, but that immediately leads to weirdness: What does it
mean for the dependency to a product if said product is enabled, but its
Export item is not? What about conflicting conditions in the Export
items of an inheritance chain -- do they get logically "and"ed? And so
on.
Change-Id: I08d7d93d30b8a53b42a317cb3391924e4666d169
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
| |
This bug fix adds quotes around arguments with spaces
or tabs in the response files. It also escapes quotes
if they're found in the argument list.
Task-number: QBS-812
Autotest-by: Joerg Bornemann <joerg.bornemann@qt.io>
Change-Id: I4f4479e413b467e64f55c3068475d0dbb1990798
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
Reviewed-by: Philip Schuchardt <vpicaver@gmail.com>
|
|
|
|
|
|
|
|
| |
qt5 always normalizes $$files()' return value
Change-Id: I06ae7bde595ae82e2ca2a25ec3269fbb877c9ae1
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
| |
Change-Id: I6589e7ebeef2dd90ae4cd82d478a8c6efc533911
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
|
|
|
| |
we don't support building with VS or Xcode anyway.
qt creator did the same move recently.
Change-Id: Iabfbe713be5954f167f0a8a4c431bd9e935c7e20
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
|
| |
We already had the respective properties, but they didn't do anything.
Change-Id: I2af86e325b6da13ea7f07eb132e207c8501ae467
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
| |
Otherwise all importing products share the same export scope in the end.
Task-number: QBS-977
Change-Id: I453dd48c17786e50e982b5e0369434b1108705cc
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
| |
Task-number: QBS-977
Change-Id: I7e5e287568b674402bb2dfeb00875da9d2620f36
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
| |
Change-Id: Id543798e8c00840504116d9077288327ab0f6543
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
|
|
|
|
|
|
|
| |
The _MSC_VER define must be passed to moc and friends.
Change-Id: I77d1a030e802a24b6980d67434e05ae0be96a20d
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
|
|
|
| |
It would sometimes fail because the actual output is potentially
interspersed with qbs status messages. Take this possibility into
account.
Change-Id: Ib1b9e05a754bd8bf56d6834decd49c10838ed0c0
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
|
|
|
|
|
| |
Change-Id: I5a42c678d17951c7da12a88683bbd32fb2ed5f4f
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
| |
Change-Id: Iaba0d59741355b70923c24e8f9bc502c3db4856e
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
They can get quite expensive, particular with null builds. Allow to re-
enable via environment variable.
Change-Id: If8368e14f97223fc0398aa1111711c4d131a2fb4
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
| |
It is true that all outputs of a transformer get the same timestamp
after it has run successfully, but a new output being added afterwards
will have no timestamp at all, so we cannot rely on that as part of the
up-to-date check.
Task-number: QBS-945
Change-Id: I5568ec9854fca46c1ca9fbd5cc95c9b1e3351c05
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
|
|
| |
It was an oversight that we did not have this before.
Change-Id: I6347be273273da301ca1ee3e23ff2ddda0a1553a
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I754280c0b8b02c4693be6936b67f458f1a57ba2b
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|