| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The '9.0.1' version was removed from the mirror we
use to install QtC.
Bump version to '11.0.3' to use newer Qbs.
Change-Id: If884d8b2cf9c1d2844ad01a80ebb37c3fc94eff9
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: Id477de4750ac3bce8e39abe5d2195c75ca0addef
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
| |
| |
| |
| |
| | |
Change-Id: I38fa9d66a8e136c498f5826e2d3d84fe06bafe89
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Ubuntu Focal is almost 4 years old now; some packages, like
LLVM, are quite old as well - it would be nice to use
the same llvm verion that macos github runners use (which is
15, the latest is 17), while only llvm-12 is available on focal.
Change-Id: I5f033abb41ab62c58e204307f04b2a224f8e59bf
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| | |
Brew clang complains about deprecated sprintf.
Change-Id: I814b2ba4bf37616ff98aeef81efd6965ee680073
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: Ib1edb43b9361f866a9ed5f2110c51945c32db5fa
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
We recorded both too many and not enough time stamps for wildcards:
- We should not try to track changes regarding recursive wildcards
("**/"), as that can easily pull in large parts of the source
tree and/or the project root directory, likely resulting in
unwanted re-resolvings.
- We should track the parent dir of a wildcard pattern, even if
the last evaluation yielded no matches.
Change-Id: I8e5521a2fcb7fd1f54103e0df566f33dd6f2a972
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
| |
| |
| |
| |
| |
| |
| | |
macos-11 is deprecated
Change-Id: I4ce2c5373e09fe1396d58ff8ffd318eb037f13c0
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Otherwise, we will potentially emit unwanted warning messages.
Amends 5af4f376f88e7859667de6be0eff8f8af304528d.
Change-Id: Icf55cfde089ae9f371c125bdad3eb79768070578
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
| |
| |
| |
| |
| |
| |
| |
| | |
PATH should be a ';'-separated string on Windows, not a
JS array.
Change-Id: If4711fc76c530376812328bced971babb26abac9
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|/
|
|
|
|
|
|
|
| |
... when resolving wildcards.
Somehow, this seemingly long-standing typo appears to have been exposed
by the refactoring in 1490cf65966007ec7564475da9d606d3ae6475fc.
Change-Id: If4e9e19476cb25bcbc966d2ca525ff0530d0b419
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
|
|
|
|
|
|
|
|
|
| |
This feature uncovered some sloppiness in our handling of QVariant types,
which has now been fixed.
Task-number: QBS-395
Change-Id: I687cef7470d97fe7887e4a7a1dbe672b2b9c79ec
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
Reviewed-by: David Schulz <david.schulz@qt.io>
|
|
|
|
|
| |
Change-Id: If069568ecae202ea5a2280ed43c42607f9433648
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|
|
|
|
|
|
|
|
|
|
| |
The fallback provider creates modules for all named dependencies that
were not found and only later checks during validation whether such a
pkg-config module exists. This breaks assumptions all over the place.
Since we have a superior replacement in qbspkgconfig, we now deprecate
the fallback provider.
Change-Id: I3694515865d63e7308cfab7778068aa2e480fe23
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Now that our CI complains about formatting, we need a simple way to fix
it beforehand. Therefore, we import this helpful script from LLVM. Copy
it into some directory in PATH and then invoke it like this to format
the last commit according to our coding style:
$ git clang-format HEAD~
The -h option gives more usage examples.
We might also want to add a pre-commit hook later.
Change-Id: I0fe775a32fc841848288ecea330bde303d9e8323
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
|
|
|
|
|
|
|
| |
This example contains two libs that are exported
using multiple exporters (both qbs and pkgconfig).
Change-Id: I67a4a2461d81be7931767e8cc0b968d05cbdf48c
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|
|
|
|
|
|
|
| |
This change adds a step-by step tutorial that helps
to explain Qbs concepts and best practices for new users.
Change-Id: I5c669f8fa0f89b8300f241bb8e4ed7cd4b3bb4c6
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|
|
|
|
| |
Change-Id: I4c52ae7b9c255b23028d1cd6c77ff4fdf4c00e42
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|
|
|
|
|
|
| |
As promised.
Change-Id: I2e13d220057225f1afcad0cab1e5e049313cae2b
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
| |
We have a dedicated test that checks that examples are built.
It makes little sense to build examples as a part of Qbs since
it prevents us from installing example binaries which makes
examples incomplete.
Change-Id: Ie14ba8bf12daa05f6de12d8a8873a704020852c2
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|\
| |
| |
| | |
Change-Id: If3a5e7af6e23d2f5d6b2c3e1e2531ba55090a625
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This prop was broken during refactoring.
This amends d99256dd79460628aafb5fa34a8dde7761ff7b1c.
Change-Id: I7932fc22e97c7af18bb269ddc2344c8745a82002
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Fix the probe by returning only paths to qmake
which are later passed to real Qmake probe which
checks if it should be run (e,g. for Qt modules).
This amends d99256dd79460628aafb5fa34a8dde7761ff7b1c.
Fixes: QBS-1779
Change-Id: Ie901befcc74713a41258ba18c436435fe50cfb4d
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: I1ebd7fd0266bc058d3bef725abb664662d749600
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Make sure that clang-format enforces 100 character limit
by setting PenaltyExcessCharacter.
Also lower the penalty for breaking before function args
so that it does not break after the first . for member
functions.
Note that current penalty values are way too high - everything
over 100 is used for "Stuff You Really Don't Want".
Change-Id: I5bff316b98237ceb22929e4ed1f26a314d31a562
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
pros:
- when renaming a function, arguments are not realigned
- can have more long/complex expressions as arguments
especially with long function names
cons:
- code huddles to the left side which migh seem weird
Change-Id: I9f1d1632dffcf29d08da8b9417d2c521096a1e9b
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| | |
Fixes: QBS-1748
Change-Id: Ib8a0fb8eaa6076934396eeb209133411d0384f15
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Previously, recursion was not terminated for private
dynamic libraries leading to slow performance.
Also, remove now redundant deduplication that happened
later.
Fixes: QBS-1714
Change-Id: Ia28f9993de8a55d6615c4ba0f7a4ef90721e31ff
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|\|
| |
| |
| | |
Change-Id: Id7e1b77e70eb04ff84699d2f32ec80f55c8b7ec0
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The algorithm did not properly handle names with more than two
components.
Fixes: QBS-1772
Change-Id: Ia680c69907c449e46977f2c5615707805a3b785f
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
We used to do that anyway and then suppress the resulting exception, but
that can lead to an unexpected state later. The sequence that triggered
the linked bug was as follows:
- load module via non-required dependency in other module
- exchange placeholder item in the product item and try to attach
non-existing property on the module instance
- suppress the resulting exception (because the dependency
was not required) and continue
- load module again (required this time) on product level
- exchange placeholder item again -> assertion, because we have a
module instance instead of the expected placeholder
Another way to fix this would have been to revert exchanging the
placeholder item after encountering an error, but that would be more
complex, and there is no reason to attach properties to non-existing
modules anyway.
Fixes: QBS-1776
Change-Id: I581a076d1d872616b186e4015873baeea211b647
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Our tests did not cover the case when there is obviously
no Qt that come from this provider.
Add this case and check that we only get the error about
missing module but no error come from the provider.
Fixes: QBS-1777
Change-Id: Id3e80fd54e1519c91be37346276fd781d9f81b0a
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|\|
| |
| |
| | |
Change-Id: I98cbee39cba1acd8a16eca1c9a11476e09011c15
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
It only fails with "Can't find cpp protobuf runtime. Make sure .pc files
are present" which is not helpful, since they are present.
Additionally remove the grpcpp to grpc++ mapping in both pkg-config
probes.
This kind of amends 1772df50a3bacd91ba8125ececf774aa0e6311f9, I readded
the tests.
Task-number: QBS-1663
Change-Id: I73b7b76a66cc2b8eea5dddd57e5221715b6e6014
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
On windows/msys, it is crucial to use autodetection for package
paths - otherwise, Qbs is not able to detect e.g. protobuf
package.
The pkg-config has the --define-prefix switch that overrides
the ${prefix} variables on the package with one based on
the .pc file locaiton. Implement similar option and set it
to true on Windows.
Also fix setting sysroot to avoid duplication of sysroot
in paths when prefix is autodetected.
Change-Id: I2754e48a96cb2e5f01ecace616188782cb5b0d7a
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Filename should be the last argument.
Fixes: QBS-1774
Change-Id: I6a45df6deacdba44bbb28d54ec8a90c2ade4525d
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: I325a3843e9b35eb92d911d3f8c6ed55976962950
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: Ifb61a6ac785ec605a3912b87b176251d342df958
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
|\|
| |
| |
| | |
Change-Id: I03789600e30aaa8d01d56777f08319d78813a260
|
| |
| |
| |
| |
| | |
Change-Id: I1527740715b362fa4ff0be176e1b813baebad557
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: I6205f1ad4955966e8dbc5a93bc94d84490c6d011
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The pkgconf in MSYS reports relative paths which we treat
as relative to the binary.
Also, on windows, ';' is used as a separator between paths.
Change-Id: I3425ce896e42f87da0cab3f8f06892e28d4e5727
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: I52622831cf9d0641ce1c24f53697baf24c891528
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
According to the ld man, this mode is deprecated; with clang
15.0.0, it is no longer possible to pass this option
via compiler frontend (it treats -lazy_library flag as
-l flag and tries to link to "azy_library").
Change-Id: I74a877c359ae3550d372486e055a9b4c3e9be997
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| | |
For some reason, Apple clang complains about missing
_main; adding single ret instruction fixes this.
Change-Id: I286f774e462e747dcff9e0c9e44b1e6a68fa432f
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| | |
We have to turn them into JS values at the API border.
Task-number: QBS-1770
Change-Id: Ia3178c9cc7611b9609cb8d006c742434d299873c
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
It's not only cumbersome having to write out the name of the surrounding
module, but it actually prevents some legitimate use cases, e.g. when
using a Module base item where the name of the derived module is not
statically known.
Task-number: QBS-1770
Change-Id: Iad4f6a16a27c51b303f26bc98fd9b2a4174d5f42
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
|\|
| |
| |
| | |
Change-Id: Ibd6fef8ad780af22bd39e583ca36f8994fb6510c
|
| |
| |
| |
| |
| | |
Change-Id: Ifb7b2f6de53db9f709952439749d89ff1234d1f9
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Microsoft's baseline is a couple of hundred years before Unix's.
Fixes: QBS-1768
Change-Id: I1845036b59665cbc1640887840a7d03edbc0cb54
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
|