| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
|
|
| |
Lighthouse was an internal name, it's now called QPA. Anyhow, this
reference adds nothing to the understanding of the section, so we can
as well remove it.
Change-Id: I92808069a70d75a0ef9741566e0f9bc3b746231d
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
| |
Saying that you need to 'import' the dependency is unfortunate,
since import has been introduced with a very distinct meaning
(QML imports).
Change-Id: I59216cb651853eafb8169966cd6b23cee3201268
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
| |
Task-number: QBS-910
Change-Id: I1f24edceeae857404874db5be2c09e7d03836e53
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
| |
Task-number: QBS-908
Change-Id: I16c2ea7ade8f3be0c1cb18acbe06c737c359cad5
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
|
| |
|
|
|
|
|
|
| |
valgrind inserts LD_PRELOAD into the environment, screwing up our null
build benchmark. Work around that.
Change-Id: I3acf2e19a9d5a1f9b2bdfdbd044a7dd8ef64574c
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Contrary to what the old code assumed, it is not enough to compare the
values of environment variables that were used while resolving the
project: A command run by a transformer might be sensitive to
environment changes as well (e.g. gcc to CPLUS_INCLUDE_PATH).
Note: This does not mean we will rebuild everything when anything in the
environment changes; it means that *if* a file is rebuilt, its command
will see the new environment instead of the old one.
Change-Id: I7916b8c40f1440564034ec3317c549ebd4e988a0
Reviewed-by: Jake Petroules <jake.petroules@theqtcompany.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
| |
Change-Id: I73586e18f256ef5f1985df2f16f353384d9b26e4
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
| |
Task-number: QBS-887
Change-Id: If6a12f42c26fa3143a1aa97947b659226ecbff03
Reviewed-by: Jake Petroules <jake.petroules@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
| |
Example output on that platform: "Native code performing complex
internal combustion process (000000000212F1F8, 000000000234F5F8)!?"
So no "0x".
Change-Id: I2527447241d9ec5d3dbee09277715e5838ad8199
Reviewed-by: Jake Petroules <jake.petroules@petroules.com>
(cherry picked from commit 419180b289bd78b13ed9f51de6998e2decb079e3)
Reviewed-by: Jake Petroules <jake.petroules@theqtcompany.com>
|
| |
|
|
|
|
|
| |
To 1.4.4.
Change-Id: Ice951253c665ba76caf290208415203a7fc6d15b
Reviewed-by: Jake Petroules <jake.petroules@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
| |
In EvaluatorScriptClass, we overwrote the already existing outer item
with a (non-working) new one. The original intention of that code is not
entirely clear to me; my assumption is that it can continue to exist as
a fallback where "outer" is used outside a group.
Task-number: QBS-877
Change-Id: Ife0b7f85574f7d880307cd2f65f551a64ca3cf6f
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
| |
Change-Id: I83b53ec2d37a2e5b499b1cd81b1cd2fb3e593750
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
| |
This example did nothing since commit c91dcd35.
Change-Id: I7aa440272b2499e26bc947e9b973682a82ca2b75
Reviewed-by: Jake Petroules <jake.petroules@theqtcompany.com>
|
| |
|
|
|
| |
Change-Id: I4115a6e2972c83f6764596b63f882f356feb8db5
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
| |
Mainly to make binding to module properties work. This should also
reduce the potential for other suprising behavior, as now all following
code sees the Group items where they always were.
Also tighten the autotest.
Task-number: QBS-848
Change-Id: Iebce3f6534a4faace5c9c8bcf4447b938b9a651b
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
| |
Was forgotten in 9d025655d5.
Change-Id: I10d37c47be99d6366bef857218d7d895e0c71a62
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The following sequence was problematic:
1) qbs build someModuleProperty:true
2) qbs resolve someModuleProperty:false
3) qbs clean
Step 2) causes the artifact objects to get removed from the build graph
and their relevant information stored in the list of rescuable artifact
data, so that on the next build the files will either get re-created as
artifacts or removed from the disk, depending on whether the changed
property causes their transformer command to change.
The problem was that the ArtifactCleaner only looked at the build graph
and was not aware of rescuable artifact data. With this patch, we go
through the list of rescuable artifacts and remove the entries from that
list as well as from the disk.
Change-Id: I05348fa1e92268c58f75074ea99371e39cc662f4
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
| |
By moving the call to loadBaseModule() before the one to
checkItemCondition().
Change-Id: I7536dcf54fbf572112ce914d6cce3c941efc1338
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
| |
Change-Id: Ic526342ba837a92dfee0f79ee37c0d8c586c200e
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
| |
|
|
|
|
|
|
| |
actool requires --output-partial-info-plist to be specified if
--app-icon is.
Change-Id: I558ff6be4f78eaafbee22726b9a00c4fdf7ba82d
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
This fixes an issue where asset catalogs compiled their resources into
the wrong directory due to the ib module reading properties of the
bundle module from the dependency in its own scope, rather than the
product's scope.
This could be triggered by, for example, setting the consoleApplication
property in a product to true (which causes bundle.isBundle to be false)
while also explicitly setting bundle.isBundle to true.
Change-Id: Ie014d42d4d8bef5e0b253c0f5a79036b4791ed55
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
| |
|
|
|
|
| |
Task-number: QBS-863
Change-Id: I4fcd08fcd8c055bb12afe28a344f00f3b2cfe9c1
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
| |
Task-number: QBS-858
Change-Id: Iafb1eb61e9e7f41fa9e5c5d1f3a8e7f58b6c1a60
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
| |
The code assumed that an absolute path on Windows always starts with a
drive letter.
Task-number: QBS-862
Change-Id: I3c673f4539dd0967a8b3164522bed51e6dfa667e
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Not just the nodes in the products we want to build.
Otherwise, in an IDE like Qt Creator the following scenario is possible:
- User does a normal build and cancels it.
- User then builds a single product.
- Now there are potentially nodes outside of this product
in state "buildable".
- In finishNode(), these get added to the list of leaf nodes
and we will erroneously visit them.
Task-number: QBS-835
Change-Id: I46d98e80cc2ebc8e88316201a56bb671af5d70de
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
| |
Change-Id: Idef447a7b2cd09a0b66d2bd785b2969a87dd51f8
Reviewed-by: Jake Petroules <jake.petroules@petroules.com>
|
| |
|
|
|
|
|
|
|
| |
Without this patch, we simply run the unknown executable without any
arguments, which can hang forever if it takes its input from stdin by
default.
Change-Id: I846821c3bb7321435b84d6fae5e26d97254b20bd
Reviewed-by: Jake Petroules <jake.petroules@petroules.com>
|
| |
|
|
|
|
|
|
| |
If the path does not contain a dot, the functions return the complete
path instead of the file name.
Change-Id: I29d1cae78375ad89ead2e467ce1c76c3e48230a1
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If we attach values to properties of a conditionally pulled-in module
inside a Properties block, that binding gets inserted into the
surrounding module. Without this patch, the property declaration check
will then emit a diagnostic for that property if the condition of the
respective Depends item is false. We fix this by skipping the check for
bindings that were not originally present in the surrounding item.
Note: This means we will not catch some bindings anymore that are
actually wrong. It could probably be done, but would involve an
inordinate amount of book-keeping, plus additional evaluations of
Properties conditions.
Task-number: QBS-854
Change-Id: I75c2af00fdfb97c9db2de2241f701dd213a401da
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
| |
These were apparently forgotten in f1db83fdfe.
Change-Id: Ia4a82f49bdfadb879b8ebca7400c7be9774b5523
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
|
| |
For example, on OS X, /tmp is a symlink to private/tmp, so attempting a
build in /tmp would generate moc_*.cpp files whose number of ../
components would be incorrect when later resolved against the
canonicalized path by the compiler, leading to build errors as a result
of the included header not being found.
Task-number: QBS-801
Change-Id: I85ce737baf77c98b860b61c27b24b7b017d9d19e
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
So far, users only got warnings for invalid property assignments in
existing modules, but neither for assignments to non-existing properties
in the Export item itself nor for those to properties in unknown
modules.
Note: The errors we already caught before will now be reported twice.
This should ideally be fixed in a follow-up commit.
Task-number: QBS-851
Change-Id: Ie00d6af8d5817bb5ecdebf6ffd735b5156b558df
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
| |
Task-number: QBS-851
Change-Id: Ice8f9d298ba65e60e2ad33dc9d5f8b2bb8961706
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
| |
Only skip these items when they occur as child items (of the exporting
product), but not when they occur as part of item values (of the
importing products).
Change-Id: Id0193c2c44250461c69357705f8703b8defa082e
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
| |
Otherwise, potentially large parts of a project will go unchecked.
Change-Id: I61b2622ae1c3e249836d75a72cd3a8145ba5ebb2
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
| |
Change-Id: Ic0a563013157343bdc2e621da38652f23de93086
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
| |
|
|
|
|
| |
Task-number: QBS-847
Change-Id: Ie4f7c27aac2ebdda6d51bd8892e0a3e8c7e826cf
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
| |
"Fix" by adding the .exe extension, and "improve" by adding the standard
install locations to the search path if node.exe is not in the PATH.
Change-Id: Icff5f1d733a663737e9d5fbddaa00b72d8d3620c
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
| |
A side effect of the bugfix commit 33c9a2721b was that two imports of
the same file were no longer considered the same if the import location
differed. This results in a considerable slowdown due to more cache
misses, so restore the old behvior in this regard.
Also, JSImport's qHash() and operator==() are now consistent.
Change-Id: I23acd665a6329ec1ab132cb34657f0b05357093f
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
If a product or higher-level module wants to set the property of a
(different) module only if some condition is fulfilled and leave it
unchanged otherwise, it would currently need to duplicate the value
from the module prototype. Obviously, this is not feasible for anything
but the most trivial definitions, and even then it easily introduces
maintenance problems.
Change-Id: If33c93a25dc84603fe95704e8815aa6e3aa1e0e4
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
| |
The form "import MyCollection" was supported, the form "import
'mycollection' was not. That does not make any sense.
Change-Id: Id57b6749300bfa21cd1afcbc58d234e5c67a963d
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
| |
No need to clone the prototype item for every single list property.
Change-Id: Iccde5172aa093ca1df25d93fbf3c336ed3445ab0
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
| |
Task-number: QBS-845
Change-Id: I885959f930d84aefa3013d29b6bef61f01fcc8cb
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
| |
The old implementation only made the contents of a single file in the
directory available (namely the last one it happened to read).
Change-Id: Ib88cb738bef161bd1c2c5ad842cf7ff33ca171b6
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
| |
Change-Id: I3d16bb9aa8b742fa37281f66ba19fc1bad5d8147
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
|
|
|
|
| |
This is more consistent with the rest of qbs and as a helpful side
effect, encourages Android profiles to be re-created since their
generation was recently completely overhauled.
Change-Id: I1041584eb197b39d9ca3634687ea86d2fb95a2d5
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|
| |
|
|
|
| |
Change-Id: I6b0397b1a8952bf6f48319c3c5d41344f12fc6ea
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
|
|
|
| |
Change-Id: I9f7bea17b70466b1e84721f3a03f723b8f3b5cae
Reviewed-by: Jake Petroules <jake.petroules@petroules.com>
|
| |
|
|
|
|
|
|
|
|
|
| |
- Remove unneeded #includes (particularly from header files).
- Make inline member function implementations less verbose.
- De-QObjectify the Loader class and remove unneeded members & heap
allocations.
- Minor stuff (const issues, unneeded namespace qualifications).
Change-Id: I55f2d7ed98b32faed7582bc6adfa04eb799de874
Reviewed-by: Jake Petroules <jake.petroules@petroules.com>
|
| |
|
|
|
|
| |
Task-number: QBS-832
Change-Id: I7b2df4ba78ac3b8417b2e0338b909641915fabf2
Reviewed-by: Christian Kandeler <christian.kandeler@theqtcompany.com>
|