@AudriusButkevicius Related to your hotfix release 0.10.13 and commit ( https://github.com/syncthing/syncthing-android/commit/a9a863ace95174d5742dc38e90280b5949dc1b79 )
First, thanks you took care about fixing the russian encoding problem by a workaround. I said, I disagreed with your commit without review to the master and pushing this immediately public.
Did you investigate this issue ( https://github.com/syncthing/syncthing-android/issues/1198 , https://github.com/syncthing/syncthing-android/pull/1199 ) deeply as I did before changing the code, honestly?
Did you know your fix is putting phone at danger to be deleted accidentially as the code now (according to my reproduction with the folder name the user in issue #1198 mentioned ) fails to create the test file and then a delete command is put into the shell with all that encoding problems? I f**** up some data on my phone yesterday investigating this issue as I had an un-deletable, wrongly encoded folder in my filesystem during testing different cyrillic folder names and also tried the “rm solution” to get rid of them as … suddenly one cyrillic name ended up in the encoding failure and rm began a file deletion in the root. I just needed time to find a solution and already said I’d fix this. I requested you to merge my workaround PR #1199 (dirty workaround not doing shell stuff on foreign language and returning true which restored UI functionality to that of earlier releases) until my proper fix is ready. I’m currently testing this and this time is really needed to don’t get any more bad. Why didn’t you just make the hotfix release containing the urgent PR about the topic of this thread? We just had one(!) user reporting the encoding issue on folder creation using cyrillic language. I can’t get it, sorry. You were the one reviewing very detailed my PR’s and yourself don’t get into review? @imsodin also offered help, so there are at least three people available taking @capi and me into account.
We really should get better! Related to your more and more “test would’ve helped” comments, yes maybe that’s true, but a human testing also has a value and both testing strategies are nothing if our release cycle is react instead of act in proper and timed order.
P.S. I’m reopening issue of the user (as you closed it and provided a workaround yourself rejecting mine with the need of a proper fix). I’m reopening my workaround PR and titling it “WorkInProgress” offering the proper fix I found some minutes ago. I’ll tell you when I’m done with testing on my phones as I’m not allowed to merge/release. Then, my suggestion is to make 0.10.14 BETA with the proper fix for the encoding-in-shell problem contained.
P.P.S. As I’m willing to invest time, things would go faster if I could act like @Nutomic did it in former times. When fix>beta>fix>beta> … cycles would be over and quality assured together with the beta users, this can of course get the “long way”. Decide if you want it. Deal is I catch up the work “Nutomic formerly had less time and didn’t do” to be compatible with modern phones Android (especially 8+ where we now have most problems with the service background ANR timeout resulting in a “crash”), take responsibility for my changes on the beta channel and offer support, make some sensible approaches to restructure existing functionality in a new UI (including user feedback) and have more flexibility in trying code out with those ~ 800 users. (Not for the master, of course!)