Re: [darktable-dev] filmic crashing dt
Something is going on, it doesn't crash for me, but if I start a new edit, disable the base curve, and in filmic use the auto tune levels I get a B&W image. With and without opencl. El dom., 27 ene. 2019 a las 22:53, David Vincent-Jones () escribió: > darktable 2.7.0+338~g5b7dd1909 > > Just started today any use of filmic crashes dt > > David > > ___ > darktable developer mailing list > to unsubscribe send a mail to > darktable-dev+unsubscr...@lists.darktable.org > > ___ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
Re: [darktable-dev] filmic crashing dt
Hi David, that's not quite enough info to get started with, sorry to say. If you could load up the core in gdb and run bt full then send that info to this mail thread, that would help - as would making the gzipped core available somewhere. hth, James On Mon, 28 Jan 2019 at 13:55, David Vincent-Jones wrote: > > Don't know if this helps: > > Thread 15 "worker res 1" received signal SIGSEGV, Segmentation fault. > [Switching to Thread 0x7fffbeffd700 (LWP 19829)] > 0x7fffbe1d3b25 in ?? () from /usr/lib/darktable/plugins/libfilmic.so > > I started to rework an image that had been completed only 2 days ago. As > soon as I changed the filmic settings the curve went flat and dt shut > down. I have not changed my basic computer settings. > > I am on Manjaro (Arch repos) > > On 2019-01-27 6:32 p.m., Patrick Shanahan wrote: > > * David Vincent-Jones [01-27-19 20:54]: > >> darktable 2.7.0+338~g5b7dd1909 > >> > >> Just started today any use of filmic crashes dt > > same build on openSUSE Tumbleweed. don't have a problem. > > > ___ > darktable developer mailing list > to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org > -- Solaris software engineer / C / Python / cyclist https://www.jmcpdotcom.com/blog https://github.com/jmcp Find me on LinkedIn @ https://www.linkedin.com/in/jamescmcpherson ___ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
Re: [darktable-dev] filmic crashing dt
Don't know if this helps: Thread 15 "worker res 1" received signal SIGSEGV, Segmentation fault. [Switching to Thread 0x7fffbeffd700 (LWP 19829)] 0x7fffbe1d3b25 in ?? () from /usr/lib/darktable/plugins/libfilmic.so I started to rework an image that had been completed only 2 days ago. As soon as I changed the filmic settings the curve went flat and dt shut down. I have not changed my basic computer settings. I am on Manjaro (Arch repos) On 2019-01-27 6:32 p.m., Patrick Shanahan wrote: * David Vincent-Jones [01-27-19 20:54]: darktable 2.7.0+338~g5b7dd1909 Just started today any use of filmic crashes dt same build on openSUSE Tumbleweed. don't have a problem. ___ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
Re: [darktable-dev] filmic crashing dt
* David Vincent-Jones [01-27-19 20:54]: > darktable 2.7.0+338~g5b7dd1909 > > Just started today any use of filmic crashes dt same build on openSUSE Tumbleweed. don't have a problem. -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.orgopenSUSE Community Memberfacebook/ptilopteri Registered Linux User #207535@ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode ___ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
[darktable-dev] filmic crashing dt
darktable 2.7.0+338~g5b7dd1909 Just started today any use of filmic crashes dt David ___ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
Re: [darktable-dev] static code analysis
On Sunday, 27 January 2019 12:18:51 CET Heiko Bauke wrote: > Hi, > > I think darktable has made a huge step forwards regarding new features > in the recent 2.6 release. For this reason I would like to focus on > code quality, performance and user experience in my future contributions. Hello Heiko, static code analysis is OK. I use csbuild in my CI and have a Coverity account. However the Coverity scan service is currentrly down. csbuild is what we use internally to do static analyis on every package which goes into the distribution, but it is open source so could be integrated into CI easily. See. e.g. here: https://gitlab.com/libssh/libssh-mirror/pipelines or https://gitlab.com/cmocka/cmocka/pipelines select a pipeline and look for the csbuild runner. It wont run Coverity for you, but clang-analyzer and cppcheck and more. Extremly good are the Sanitizers. There is AddressSanitzer, MemorySanitizer, UndefinedSanitizer and ThreadSanitizer. However they do runtime detection and for that you need tests. Darktable doesn't have any testing at all. It means you would need to run darktable and edit images with it. The sanitizers make the code much slower, even with optimization turned on. Might still be worth to try them. If you write multithreaded code you should always run your stuff with helgrind and drd. Also compilers have really good warnings in the meantime you just need to turn them on and make them errors for developers. See the projects mentioned above. Cheers, Andreas ___ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
Re: [darktable-dev] static code analysis
Hi! > Doing the cleanups for this is a great opportunity for someone like me > who wants to give something back and doesn't have the time to take on a > major project. It isn't glamorous work, but I'd be happy to do it. Nice to see this happening! We already had done some static analysis but another round cannot hurt! Go ahead. Thanks, -- Pascal Obry / Magny Les Hameaux (78) The best way to travel is by means of imagination http://www.obry.net gpg --keyserver keys.gnupg.net --recv-key F949BD3B ___ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
Re: [darktable-dev] static code analysis
On 1/27/19 6:18 AM, Heiko Bauke wrote: Currently, there is an offer for open source developers to get a free license for the PVS-Studio Analyzer tool. I got one and applied the tool to the darktable master branch. ... I was not yet able to study the results in detail. There might be a lot of false positives or just minor issues. But I expect to find also more serious things. I did a survey of the report index, dove into a few dozen of the errors and found it to be a high-quality report with little in the way of false positives. A large percentage of the warnings are related to assumptions that pointers will not be NULL and a large percentage of those are directly or indirectly related to unchecked returns from malloc() and calloc(). That could be made to go away by writing wrapper functions that check what's returned and halt the program nicely in the rare event of a failure. Knowing it was that would send the developers on fewer goose chases to find the cause of a SIGSEGV further down. If performance is a concern (not that malloc() is a real screamer to begin with), the solution can be split in a way that makes it attractive for the optimizer to inline the check. Those aside, the others I looked at seem legitimate and are worth fixing. None of it will require major work. For example, https://rabauke.github.io/darktable_analyze/sources/collection.c_4.html#ln144 looks very fishy to me. That's definitely code I'd kick back during a review with a recommendation that it be written into a small function because the same logic is used repeatedly and that variable assignment inside the condition is ugly. You might get a pull request for that shortly. ;-) If I can offer a few additional comments: Before committing the project to PVS-Studio, it would be worth evaluating some of the alternatives, especially those that are open-source. I think it's great that PVS offers a free license for so many situations, but there is the risk of having to go through and re-flag all of the spots in the code where warnings were suppressed should they change the license terms or go out of business and not release the sources. Once the code is to a point where the analyzer has nothing to squawk about, static analysis needs to be repeated regularly. This could be done as a simple cron job that notifies the developers when something crops up or as a check run web hook to prevent code that doesn't pass from being committed to GitHub. I have a system with cycles and space to play that role if needed and can make whatever configuration and scripts I develop part of the dt sources so others can run it. Doing the cleanups for this is a great opportunity for someone like me who wants to give something back and doesn't have the time to take on a major project. It isn't glamorous work, but I'd be happy to do it. --Mark ___ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
Re: [darktable-dev] static code analysis
Hi there, sonarqube has some free service for opensource projects with features as of the 'developer' paid plan. (namely: scanning c++) https://sonarcloud.io/about/sq --> github integration kind regards Wolfgang On 27.01.19 12:18, Heiko Bauke wrote: Hi, I think darktable has made a huge step forwards regarding new features in the recent 2.6 release. For this reason I would like to focus on code quality, ... +1! ___ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
[darktable-dev] static code analysis
Hi, I think darktable has made a huge step forwards regarding new features in the recent 2.6 release. For this reason I would like to focus on code quality, performance and user experience in my future contributions. Currently, there is an offer for open source developers to get a free license for the PVS-Studio Analyzer tool. I got one and applied the tool to the darktable master branch. Results can be found under https://rabauke.github.io/darktable_analyze/ and are worth to have a look at. I was not yet able to study the results in detail. There might be a lot of false positives or just minor issues. But I expect to find also more serious things. For example, https://rabauke.github.io/darktable_analyze/sources/collection.c_4.html#ln144 looks very fishy to me. I hope these results help to find and to fix hidden bugs and code smells. I will update the static code analysis results from time to time. Heiko -- -- Number Crunch Blog @ https://www.numbercrunch.de -- Cluster Computing @ https://www.clustercomputing.de -- Social Networking @ https://www.researchgate.net/profile/Heiko_Bauke ___ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
