Feed Advanced Search

Thu, Jan 18

mwolff committed R45:232c92992d19: Fix bounds check: 0 is a valid index (authored by mwolff).
Fix bounds check: 0 is a valid index
Thu, Jan 18, 10:47 AM
mwolff committed R45:1412055f818f: Also override relocations in DT_REL (authored by mwolff).
Also override relocations in DT_REL
Thu, Jan 18, 10:47 AM
mwolff abandoned D9766: Use at least the requested width for the argument hint tree.

this was comitted already, closing it again

Thu, Jan 18, 10:29 AM · Frameworks, Kate
mwolff committed R244:16e8cf1e6607: Optimize: use QMetaObject::invokeMethod with functor (authored by mwolff).
Optimize: use QMetaObject::invokeMethod with functor
Thu, Jan 18, 10:17 AM
mwolff closed D9823: Optimize: use QMetaObject::invokeMethod with functor.
Thu, Jan 18, 10:17 AM · Frameworks
mwolff committed R244:421e85344db7: Add benchmarks for KDirWatch (authored by mwolff).
Add benchmarks for KDirWatch
Thu, Jan 18, 10:17 AM
mwolff committed R244:5d91ecc2fb68: Use QSignalSpy::wait instead of QTest::wait where possible (authored by mwolff).
Use QSignalSpy::wait instead of QTest::wait where possible
Thu, Jan 18, 10:17 AM
mwolff closed D9821: Use QSignalSpy::wait instead of QTest::wait where possible.
Thu, Jan 18, 10:17 AM · Frameworks
mwolff closed D9822: Add benchmarks for KDirWatch.
Thu, Jan 18, 10:17 AM · Frameworks
mwolff committed R244:311a702fa070: Verify that the file was opened (authored by mwolff).
Verify that the file was opened
Thu, Jan 18, 10:17 AM
mwolff committed R244:5fa1aaa2a47b: Verify that the path is valid and writable (authored by mwolff).
Verify that the path is valid and writable
Thu, Jan 18, 10:17 AM
mwolff closed D9820: Verify that the file was opened.
Thu, Jan 18, 10:17 AM · Frameworks
mwolff closed D9819: Verify that the path is valid and writable.
Thu, Jan 18, 10:17 AM · Frameworks
mwolff accepted D9956: Fix SVN history fetching and clean up plugin.

I'm OK with this, but please split this up the next time. You could still have posted it as one review request, but splitting it up is always a good idea. It's a general coding mantra that you should start following, as it makes your life much easier in the long term.

Thu, Jan 18, 10:09 AM

Wed, Jan 17

mwolff added a comment to D9259: fix heaptrack_inject to find second kind of relocation.

I've adapted and pushed this now. Many thanks Ivan! Much appreciated.

Wed, Jan 17, 3:40 PM · Heaptrack
mwolff committed R45:ab2a8222da3c: Check symbol table size before accessing it (authored by mwolff).
Check symbol table size before accessing it
Wed, Jan 17, 3:39 PM
mwolff committed R45:fe3f4d769d90: Make run_linkage_test.sh pass (authored by mwolff).
Make run_linkage_test.sh pass
Wed, Jan 17, 3:39 PM
mwolff committed R45:56140865b28a: Support --debug/-d switch when attaching to a process (authored by mwolff).
Support --debug/-d switch when attaching to a process
Wed, Jan 17, 3:39 PM
mwolff committed R45:779dceb8cbe7: Override allocation functions accessible through .rel{a,}.dyn (authored by mwolff).
Override allocation functions accessible through .rel{a,}.dyn
Wed, Jan 17, 3:39 PM
This revision was not accepted when it landed; it landed in state Needs Review.
Wed, Jan 17, 3:39 PM · Heaptrack
mwolff committed R45:943eb7ab02bd: Check symbol table size before accessing it (authored by mwolff).
Check symbol table size before accessing it
Wed, Jan 17, 3:18 PM
mwolff committed R45:5f9a61d06a89: Support --debug/-d switch when attaching to a process (authored by mwolff).
Support --debug/-d switch when attaching to a process
Wed, Jan 17, 3:11 PM
mwolff committed R45:0ec8d2c49dc6: Override allocation functions accessible through .rel{a,}.dyn (authored by mwolff).
Override allocation functions accessible through .rel{a,}.dyn
Wed, Jan 17, 2:59 PM
mwolff committed R45:7af4b27646de: Override allocation functions accessible through .rel{a,}.dyn (authored by mwolff).
Override allocation functions accessible through .rel{a,}.dyn
Wed, Jan 17, 2:57 PM
mwolff added a comment to D9259: fix heaptrack_inject to find second kind of relocation.

Aha, -DTAKE_ADDR fails when I use ld.bfd instead of ld.gold, I assume you do that too? I still don't see a difference with using -Wl,-z,now though...

Wed, Jan 17, 2:18 PM · Heaptrack
mwolff added a comment to D9259: fix heaptrack_inject to find second kind of relocation.

I fail to reproduce the issue with the provided test case on ArchLinux. When I compile the three variants, I always properly detect the free calls.

Wed, Jan 17, 1:46 PM · Heaptrack
mwolff added a comment to D9259: fix heaptrack_inject to find second kind of relocation.

Thanks, I'll take over then. What email address should I use to attribute (parts of) the patch to?

Wed, Jan 17, 12:45 PM · Heaptrack

Tue, Jan 16

mwolff committed R39:0d0284e8f8d2: Use at least the requested width for the argument hint tree (authored by mwolff).
Use at least the requested width for the argument hint tree
Tue, Jan 16, 2:15 PM
This revision was not accepted when it landed; it landed in state Needs Review.
Tue, Jan 16, 2:15 PM · Frameworks, Kate
mwolff committed R32:eeb8c640deb5: Unbreak the GDB QUrl pretty printer test (authored by mwolff).
Unbreak the GDB QUrl pretty printer test
Tue, Jan 16, 2:13 PM
mwolff closed D9922: Unbreak the GDB QUrl pretty printer test.
Tue, Jan 16, 2:13 PM
mwolff committed R32:831cf5f98d78: Link against KF5::Purpose if it's available (authored by mwolff).
Link against KF5::Purpose if it's available
Tue, Jan 16, 2:13 PM
mwolff closed D9921: Link against KF5::Purpose if it's available.
Tue, Jan 16, 2:13 PM
mwolff committed R32:a6093e7ed7b3: Also unbreak QtPrintersTest::testQByteArray (authored by mwolff).
Also unbreak QtPrintersTest::testQByteArray
Tue, Jan 16, 2:13 PM
mwolff committed R32:f4dd6425f343: Unbreak QtPrintersTest::testQString (authored by mwolff).
Unbreak QtPrintersTest::testQString
Tue, Jan 16, 2:13 PM
mwolff closed D9924: Also unbreak QtPrintersTest::testQByteArray.
Tue, Jan 16, 2:13 PM
mwolff committed R32:c08cd8060f4b: GDB Qt pretty printers: print size of container (authored by mwolff).
GDB Qt pretty printers: print size of container
Tue, Jan 16, 2:13 PM
mwolff closed D9923: Unbreak QtPrintersTest::testQString.
Tue, Jan 16, 2:13 PM
mwolff closed D9918: GDB Qt pretty printers: print size of container.
Tue, Jan 16, 2:13 PM
mwolff added a comment to D9824: Optimize inotify KDirWatch backend: map inotify wd to Entry.
In D9824#191748, @rjvbb wrote:

So please do measure the impact of your fix on performance with the QFSW backend.

Tue, Jan 16, 12:18 PM · Frameworks
mwolff added a comment to D9824: Optimize inotify KDirWatch backend: map inotify wd to Entry.

Yeah, I agree. Let's fix it. But I won't write mega patches like you seem to prefer. I try to keep things as minimal as possible. And I also can't give you an ETA or guarantee on when I'll fix the rest. This patch solves one big issues, and the added benchmark lies the foundation for future work. Let's start from here and get going.

Tue, Jan 16, 11:02 AM · Frameworks
mwolff updated subscribers of D9766: Use at least the requested width for the argument hint tree.

ping? @kfunk maybe?

Tue, Jan 16, 9:03 AM · Frameworks, Kate
mwolff added a comment to D9824: Optimize inotify KDirWatch backend: map inotify wd to Entry.

@rjvbb why did you request changes to proceed with this patch? The fact that there are more issues in KDirWatch does not mean we should hold up this approach.

Tue, Jan 16, 8:41 AM · Frameworks
mwolff updated the diff for D9824: Optimize inotify KDirWatch backend: map inotify wd to Entry.

use QHash::insert instead of operator[]

Tue, Jan 16, 8:38 AM · Frameworks
mwolff committed R299:86954e039463: Make kdesu work when PWD is /usr/bin (authored by mwolff).
Make kdesu work when PWD is /usr/bin
Tue, Jan 16, 8:36 AM
mwolff closed D9888: Make kdesu work when PWD is /usr/bin.
Tue, Jan 16, 8:36 AM · Frameworks

Mon, Jan 15

mwolff updated the diff for D9888: Make kdesu work when PWD is /usr/bin.

cleanup now that I figured out what happens, thanks @sitter

Mon, Jan 15, 1:14 PM · Frameworks
mwolff added a comment to D9888: Make kdesu work when PWD is /usr/bin.
[13:35] <milian> how the hell can kdesu ever work? https://phabricator.kde.org/D9888 WTF
[13:35] <milian> does anyone know if I'm missing something?
[13:35] <milian> googling for these macros doesn't show me anything obvious either
[13:53] <sitter> milian: dead code I'd say. note the `if` following your change checks if the file 'false' is exectuable (which it isn't as that'd be ./false which likely never exists) and then falls back to QSP::findExec(cmd). where cmd is macro'd to sudo or su depending on the cmake switch KDESU_USE_SUDO_DEFAULT
[13:54] <milian> but it doesn't work for me without this patch
[13:54] <milian> so it cannot be dead code :D
[13:55] <milian> i.e. how can this work for anyone right now?
[13:58] <sitter> milian: http://paste.debian.net/1005279/ is how I read the original
[13:59] <milian> I'll see why this does not happen on my system
[14:02] <sitter> milian: try this for good measure http://paste.debian.net/1005280/
[14:02] <milian> yep
[14:03] <sitter> mind you, it could be that __PATH_SU/SUDO is actually defined somewhere in a system level include which would then make the QT_ACCESS pass and break things
[14:03] <milian> I can't find it in my  /usr/include at least
[14:06] <milian> sitter: QT_ACCESS("false", X_OK) == 0 on my system
[14:07] <milian> and no, there's no false in the current PWD
[14:07] <milian> it's in /usr/bin though
[14:07] <milian> I'd also be OK with removing that whole code and simplifying it
[14:08] <milian> like you proposed
Mon, Jan 15, 1:08 PM · Frameworks
mwolff updated the summary of D9888: Make kdesu work when PWD is /usr/bin.
Mon, Jan 15, 12:35 PM · Frameworks
mwolff accepted D9770: Code de-duplication in byteSize(double size).
  • do not ever profile a debug build, the results are completely bogus
  • do not use callgrind, use perf
Mon, Jan 15, 9:50 AM · Frameworks
mwolff updated the diff for D9822: Add benchmarks for KDirWatch.

format results on phab

Mon, Jan 15, 9:08 AM · Frameworks
mwolff updated the diff for D9822: Add benchmarks for KDirWatch.

fixup commit message

Mon, Jan 15, 9:07 AM · Frameworks
mwolff added a comment to D9824: Optimize inotify KDirWatch backend: map inotify wd to Entry.

150KB is not a lot of memory

Mon, Jan 15, 9:04 AM · Frameworks
mwolff added inline comments to D9821: Use QSignalSpy::wait instead of QTest::wait where possible.
Mon, Jan 15, 9:03 AM · Frameworks
mwolff updated the diff for D9823: Optimize: use QMetaObject::invokeMethod with functor.

thanks dfaure ;-)

Mon, Jan 15, 8:40 AM · Frameworks

Thu, Jan 11

mwolff added a comment to D9823: Optimize: use QMetaObject::invokeMethod with functor.

https://codereview.qt-project.org/#/c/216497/

Thu, Jan 11, 4:52 PM · Frameworks
mwolff accepted D9809: work around bug in kLineEdit.

hm ok then

Thu, Jan 11, 4:22 PM · KDevelop
mwolff added a comment to D9823: Optimize: use QMetaObject::invokeMethod with functor.

cleaner, yes. but also much slower. contrary to the other code-paths, the QTimer::singleShot taking a functor is not optimized (yet?) for timeout == 0...

Thu, Jan 11, 4:08 PM · Frameworks
mwolff requested changes to D9809: work around bug in kLineEdit.

So just remove the assert - instead return early when indexes.isEmpty(). That way we don't need to mess with the signal blocker and the code behaves as intended. And you remove code instead of adding more.

Thu, Jan 11, 3:55 PM · KDevelop
mwolff updated the diff for D9823: Optimize: use QMetaObject::invokeMethod with functor.

make compile against older Qt

Thu, Jan 11, 3:52 PM · Frameworks
mwolff planned changes to D9823: Optimize: use QMetaObject::invokeMethod with functor.

requires 5.10, so I can't commit this as-is...

Thu, Jan 11, 3:44 PM · Frameworks
mwolff added a comment to D9824: Optimize inotify KDirWatch backend: map inotify wd to Entry.

@rjvbb this patch just shows that KDirWatch has tons of performance issues that need to be fixed here. Please do run the new benchmarks on your benchmark with your backend, profile them, optimize them.

Thu, Jan 11, 3:14 PM · Frameworks
mwolff added reviewers for D9824: Optimize inotify KDirWatch backend: map inotify wd to Entry: rjvbb, KDevelop.
Thu, Jan 11, 3:13 PM · Frameworks
mwolff accepted D9812: [Icon Item] Treat sources starting with a slash as local file.

lgtm

Thu, Jan 11, 10:24 AM · Frameworks, Plasma
mwolff requested changes to D9812: [Icon Item] Treat sources starting with a slash as local file.

lgtm in general, but can be cleaned up

Thu, Jan 11, 10:00 AM · Frameworks, Plasma
mwolff added a comment to D9770: Code de-duplication in byteSize(double size).

10 calls per second sound fine to me, that shouldn't be a big performance issue at all. Are you measuring performance of a debug build or of a release build? Can you specify the exact commands you are profiling? Is the performance better when you are using KFormat here?

Thu, Jan 11, 9:35 AM · Frameworks
mwolff added a comment to D9809: work around bug in kLineEdit.

I don't understand how this fixes the crash. Can you extend the bug report with a better backtrace that contains more symbols? Also, the bug report looks like it's due to a queued signal emission. Which signal is that? And it looks like something withing the kdevplatform project library, how is that related to the compiler selection?

Thu, Jan 11, 9:02 AM · KDevelop
mwolff added inline comments to D9808: fix incorrect emission of signals by kLineEdit.
Thu, Jan 11, 9:00 AM · Frameworks, KDevelop

Wed, Jan 10

mwolff committed R52:e6dbb39f269a: Fix compilation, remove dead code (authored by mwolff).
Fix compilation, remove dead code
Wed, Jan 10, 2:31 PM
mwolff committed R32:da4268d33101: Merge branch '5.2' (authored by mwolff).
Merge branch '5.2'
Wed, Jan 10, 10:37 AM
mwolff committed R32:94ed49aceac4: Performance: Reuse the global icon loader (authored by mwolff).
Performance: Reuse the global icon loader
Wed, Jan 10, 10:37 AM
mwolff closed D9783: Performance: Reuse the global icon loader.
Wed, Jan 10, 10:37 AM
mwolff committed R32:f2a6941e086c: Cache ProblemPointers per translation unit (authored by mwolff).
Cache ProblemPointers per translation unit
Wed, Jan 10, 10:37 AM
mwolff closed D9772: Cache ProblemPointers per translation unit.
Wed, Jan 10, 10:37 AM
mwolff requested changes to D9770: Code de-duplication in byteSize(double size).

Imo this should be using KFormat::formatByteSize. https://api.kde.org/frameworks/kcoreaddons/html/classKFormat.html#ae7412420b70e2ca935d0ebed6770e313

Wed, Jan 10, 7:59 AM · Frameworks

Tue, Jan 9

mwolff added a reviewer for D9772: Cache ProblemPointers per translation unit: KDevelop.
Tue, Jan 9, 6:57 PM
mwolff added reviewers for D9766: Use at least the requested width for the argument hint tree: KDevelop, Kate.
Tue, Jan 9, 4:12 PM · Frameworks, Kate
mwolff committed R39:7c648c89ed11: ExpandingWidgetModel: find the right-most column based on location (authored by mwolff).
ExpandingWidgetModel: find the right-most column based on location
Tue, Jan 9, 10:00 AM
This revision was not accepted when it landed; it landed in state Needs Review.
Tue, Jan 9, 10:00 AM · Frameworks, Kate
mwolff committed R39:181c6146d399: Simplify code: return early to reduce indentation depth (authored by mwolff).
Simplify code: return early to reduce indentation depth
Tue, Jan 9, 10:00 AM
mwolff closed D9749: Simplify code: return early to reduce indentation depth.
Tue, Jan 9, 10:00 AM · Frameworks, Kate
mwolff committed R32:b5c7d59aad12: Add Q_OBJECT macro to ClangNavigationWidget (authored by mwolff).
Add Q_OBJECT macro to ClangNavigationWidget
Tue, Jan 9, 9:56 AM
mwolff committed R32:b4da5b152462: Remove dead code (authored by mwolff).
Remove dead code
Tue, Jan 9, 9:56 AM
mwolff committed R32:113deb25ac77: Workaround: find declarations for constructor argument hints (authored by mwolff).
Workaround: find declarations for constructor argument hints
Tue, Jan 9, 9:56 AM
mwolff closed D9745: Workaround: find declarations for constructor argument hints.
Tue, Jan 9, 9:56 AM
mwolff closed D9747: Remove dead code.
Tue, Jan 9, 9:56 AM
mwolff committed R32:67b8f6c20cdc: Further extend the argument hint code completion test coverage (authored by mwolff).
Further extend the argument hint code completion test coverage
Tue, Jan 9, 9:56 AM
mwolff committed R32:203b02d480ff: clang: Improve argument hint code completion (authored by mwolff).
clang: Improve argument hint code completion
Tue, Jan 9, 9:56 AM
mwolff closed D9744: Further extend the argument hint code completion test coverage.
Tue, Jan 9, 9:56 AM
mwolff committed R32:77d1f9d2c23d: Extend test coverage for argument hint code completion (authored by mwolff).
Extend test coverage for argument hint code completion
Tue, Jan 9, 9:56 AM
mwolff closed D9743: Extend test coverage for argument hint code completion.
Tue, Jan 9, 9:56 AM
mwolff closed D9725: clang: Improve argument hint code completion.
Tue, Jan 9, 9:56 AM
mwolff committed R32:2ad241710c46: Fix: correct order of QCOMPARE(actual, expected) (authored by mwolff).
Fix: correct order of QCOMPARE(actual, expected)
Tue, Jan 9, 9:56 AM
mwolff closed D9724: Fix: correct order of QCOMPARE(actual, expected).
Tue, Jan 9, 9:56 AM
mwolff committed R32:3e1d422eb05f: Resume background parser when replacing TestCore's project controller (authored by mwolff).
Resume background parser when replacing TestCore's project controller
Tue, Jan 9, 9:56 AM
mwolff committed R32:28c6e5d5f8d2: Unbreak most of the clang codecompletion test by passing file contents (authored by mwolff).
Unbreak most of the clang codecompletion test by passing file contents
Tue, Jan 9, 9:56 AM
mwolff added a comment to D9750: ExpandingWidgetModel: find the right-most column based on location.

I'll push it then

Tue, Jan 9, 9:56 AM · Frameworks, Kate
mwolff closed D9722: Resume background parser when replacing TestCore's project controller.
Tue, Jan 9, 9:55 AM
mwolff closed D9723: Unbreak most of the clang codecompletion test by passing file contents.
Tue, Jan 9, 9:55 AM
mwolff committed R32:f1f23ce13022: Cleanup: use auto and remove explicit KDevelop:: qualification from identifiers (authored by mwolff).
Cleanup: use auto and remove explicit KDevelop:: qualification from identifiers
Tue, Jan 9, 9:55 AM