- Queries
- All Stories
- Search
- Advanced Search
Advanced Search
Apr 3 2018
- remove old connect
- support directories again
Apr 2 2018
- TODO: new mimetype detection breaks the directory case, as it is considered an error (though we still get inode/directory as the result). Add special detection for dirs.
- do not depend on phonon to detect if we support a video format
- directly conect to mimetype signal, which appears to be emitted significantly earlier than result
- video preview resizes correctly now (for some reason videos take very long to play though)
- Text preview allows scrolling again
@bcooksley Would it be possible to set up a project page for KLook? Having all the todos as actionable tasks in a Workboard and being able to prioritize them would be more helpful than having half of them as a list in a text document, and the other half merely existing in my head. Also, when @ngraham decides to list all the usability and design improvements that are needed (which isn't really my area of expertise), he could put them there, too.
- fix initial image selection
- start porting away for KDE4LibsSupport to ease fixing some bugs
Apr 1 2018
@ngraham : So the dolphin integration works partially (as described above in the summary). I guess some input from the dolphin team would be helpful now. D11875 contains the dolphin patch so that it can be easier reviewed. I'm not so sure whether and how the discussion should be split between the two review requests.
- updated dolphin patch again
- dolphin patch applies again against dolphin master (4913493551bd4472d5a4b94397a9a21a787fe8a4)
Anything I can do to convince you that templates are awesome, and we should use them ;-)? My main argument is that this improves type safety and allows to catch off by one errors when indexing the arrays (and it enables iteration with range based for, which I haven't implemented yet, though). It also helped me catch that Q_ASSERT weirdness in D11814, as with the std::array, that assignment became illegal.
In D11779#238258, @aacid wrote:I'd say you can use override instead of Q_DECL_OVERRIDE if you want
Mar 31 2018
- split monolithic dolphin patch, currently doe only partially apply
- reduce CMake version requirement
- add .arcconfig file
In D11844#237793, @ngraham wrote:
@ngraham : I've added you as a reviewer, as I've initially stumbled (again) over KLook due to Bug 272539, and you were the last one active there. If you know anyone else who this might concern, please add them. This is really mostly meant as a RFC, and to probe if there is even interest in reviving KLook.
Mar 30 2018
In the same vein as #D11779, this is meant to modernize and clean up to make the code base more maintainable.
Mar 29 2018
In D11768#236349, @ngraham wrote:In Neon and Kubuntu, ~/Desktop exists for a new user by default. Is this not the case in other distros?
- simplify check for nullptr in assert
@aacid : You were joking that I've become the de-facto maintainer of the solver code by contributing to it, so I thought I might as well improve the general code hygiene. This patch set mainly removes provably dead code, and applies a few clang-tidy modernizations. The remainder is a bit of code simplification (reducing variable scope, malloc + memset -> calloc). I couldn't figure out how to get arc to create a series of smaller review requests (one per commit, history.immutable = true doesn't help in that regard), and hope you don't mind the clustered review request.
Mar 21 2018
Friendly ping
Mar 16 2018
Just chiming in: Are you sure this is the right place for this? Shouldn't that be done at the place where the string is displayed by using QFont::setCapitalization?
Mar 4 2018
Mar 3 2018
So the last change fixes that mp gets incremented and we happily write into the storage space of the MemoryManager which leads to all kinds of crashes. Now, I know that goto is normally not something you put into modern C++, but it seemed to be the nicest way out of the nested loop while keeping the check local to the increment.
- Don't derefernce stack if empty
- Don't attempt to store more moves than we have space for
No, I'm not sure, it was probably more wishful thinking. Though those illegal reads can cause all kinds of memory corruption. Did your crash occur in a Debug build or in a RelWithDeb build? Just asking as the first makes it more likely to be an easy to find logic error, while the latter can be the compiler making use of UB at any place.
remove whitespace change
Mar 2 2018
Similar issues also exist in mod3solver.cpp, I've added some fixes/workarounds for that one too. I hope you don't mind, else I can separate it into a new review request.
I'll probably won't contribute to Elisa regularly or any new features, but I might provide a few patches like this one to fix things clazy and other static analyzers complain about.
Mar 1 2018
Feb 28 2018
I always got the an asan abort due to the missing checks (the solver would always execute the code and dereference non-allocated memory ). Now, without asan and without the patches, I didn't get any crashes in a debug build, and occasionally crashes with RelWithDeb. I didn't get any crashes with RelWithDeb and my patches, though of course there could be other crab sources.
Feb 27 2018
I should add that KPatience relies on the Wp pointing one field before an array (which happens for instance Solver::unpack_position when strecpy returns 0, as the pile is empty). This apparently works with most compilers, but technically it's UB according to §5.7, 4 of the C++ standard (N4296). However, fixing this would require a large amount of refactoring.
This solves maybe the issue which appeared in BUG: 389540. At least, I don't get any invalid access warnings from asan anymore. However, the logic of the search is somewhat complex, so I'm not 100% sure that the correct branch is taken with the new access checks. Someone more familiar with the code might want to take a look at it.