It was actually Boris’ video that showed the strange behaviour, which prompted me to find the bug and fix it.
The point of this reply is not to boost myself but to point out that yes, there are folks involved with dt who can dive into the more complex mathematical stuff. I believe it’s important, since it’s kind of hard to find and fix bugs from one’s own code. It’s very hard to work on such complex stuff with just one pair of eyes. There have been also other people fixing issues in the newer modules.
I mean that people abuse the disconnect that being online offers so not at all what you are taking from my intension … and it could be what I said but its a great example of something comes across 180 from what was the actual intension.
I do not want to do a character assassination of anybody here. I appreciate that putting a creative group of people together to agree on the design of a piece of FOSS must be very challenging. I feel AP came across in his videos as a very determined and opinionated (not a negative comment against him) developer. He appeared to have some very black and white ideas about what was good and what was bad in DT. I remember him making a comment once where he said photography was about the look and I 100% agree with this comment. As a single example of many instances he was very critical of the shadow and highlights module and would happy tell you it was bad and should not be used, his fantastic tone equalizer module is not despite what he may claim a direct replacement of the shadow and highlights module. They produce different looks. On the surface it may appear they both brighten the shadows and darken the highlights, but S&H modules lifts the colours and the contrast in the shadows and I often like that look. It does it with a single click to activate the module. My point being that what a single developer wants to call bloat another person will call an essential tool.
I downloaded Ansel to see if it was more stable on my machine as I was having numerous crashes in DT when trying to use Astrodenoise. I have no definitive answer to if it is more stable, but what I noticed was just how much was missing from Ansel that included features I loved. It was a stripped down version of DT and for that reason I remain a dedicated DT user. I hope that AP can share and contribute his great features with DT while maintaining his personal fork. After all there is Rawtherapee and Art. They seem to co-exist without animosity.
As for comments that the GUI was bloated with too many options and too many options in the preferences, I would say there is not enough. The ability to find images using the selection filters is great, but in my opinion could go even further. As for preferences to customise DT they are great, but could have more. The only bloat I see is the bloat that makes the developer’s job impossible. So I appreciate that one day certain modules have to disappear because they can’t be maintained or consume the developer’s time for very little reward.
The tests takes too much time to be run on CI. But I do run them on one of my machine every night and the results are published into the Git integration sub-module. Today we have 136 tests and we are all green.
Yesterday I had a quick look at Ansel. I think I like a lot of what aurelien has done with the gui. Of course, it’s a pity that new cool modules such as Sigmoid and the new highlights reconstruction are missing. If I could I would create a fork of Ansel that includes new darktable features. Maybe I will try that.
Let’s try to see things objectively and without emotions, even though we must deal with negative emotional reactions. Let’s try to just see the facts and form an opinion. Not react back with emotions.
I don’t want to defend aurelien but I think I understand him. He has put a lot of work into darktable and he sacrificed a lot. Maybe secretly he dreamed of world domination or something like that. But now after several years there are not more darktable users that are real artists but most of them still have a phd in it or engineering. Of course he blames the other devs because he can’t beat up and force all those stupid photographers to use darktable. I think I see a way for a “compromise” and peace between the devs.
darktable keeps a database of edits, and it could be used to gleam information.
You could easily answer:
which modules are used in editing, and how frequently?
how many photos are actually edited (versus the whole collection)?
how often are masks used? what type?
which parts of modules are being used? (for example: Filmic has a lot of extra stuff)
which modules are used over time?
are people using keywords a lot? how many per image?
are people using saved searches? what types? (like, based on tags, based on dates, based on cameras, etc.)
how many people are using deprecated modules in new edits? (and which ones?)
how many photos are in imported into someone’s database?
which cameras are being used?
what file formats are there? (various raw formats, dng, jpeg, tiff, avif, etc.)
…and so many other things that might be useful to know
For anything to do with modules, you’ll probably want to filter by when the module stack was last edited, else you’d get a lot of old data skewing the results (especially where defaults have changed). It would be interesting to track module usage over time (probably based on either first edit, if that data is available, or last edit). You could also filter by filmic versions that are used (or if basecurve or sigmoid are used instead); some of the process versions are basically snapshots in time too.
Anyway, my point is that we don’t need to actively track how someone works; darktable already keeps all the most interesting data. And it could be done with a side-program (not in darktable itself), even as a script that queries the database and be completely opt-in. (And it should try to anonymize things, of course.)
It’s usually a lot more interesting when you have data generated by how someone works than asking them questions that they might not answer with enough detail… or their memory might unintentionally skew things. It’s best having both data and a survey, of course.