9 Jun 2020 |
Bubu | need to teach it to verify checksums though. | 21:02:45 |
_hc | glennmen: "hide results that don't have a correct localization for current user setting" will mean for many users the screen is empty | 21:10:02 |
_hc | Bubu: wow that's awesome. That'll be a really massive improvement | 21:10:42 |
_hc | I justed purged all traces of non-.yml from the fdroidserver test suite s!761 | 21:11:23 |
[gibot] | [server] !761: move tests to YAML - https://gitlab.com/fdroid/fdroidserver/merge_requests/761 | 21:11:27 |
_hc | and on that note, good night! | 21:11:44 |
glennmen | Als antwoord op @bubu:bubu1.eu need to teach it to verify checksums though.
glennmen: "hide results that don't have a correct localization for current user setting" will mean for many users the screen is empty
But they would still be shown if they have at least English, I assumed this would almost always be available or am I wrong to assume this? I wanted to improve the "quality" of results in the Lastest tab | 21:17:38 |
10 Jun 2020 |
_hc | glennmen: I think it is a correct assumption that en_US is almost always available. But it would be totally valid to have an app that only has ru for example. Like an app for Russian universities. So its not guaranteed there is always English there | 07:39:06 |
_hc | you could test to see if I'm wrong. Configure a device with only a single language that is not any form of English. It should be a language that has no translations in fdroiddata, the main app repo: https://hosted.weblate.org/projects/f-droid/fdroiddata/ | 07:41:38 |
_hc | then default F-Droid should still show apps in the Latest tab | 07:42:19 |
_hc | and none of them should be in the device's language, to be sure that this test was properly setup | 07:42:41 |
_hc | I choose languages that do not use a latin alphabet for tests like this, so its obvious | 07:43:13 |
glennmen | with my change this will still show results even if your setup your device in a language that doesn't even exist yet in fdroiddata | 07:58:26 |
glennmen | the Latest tab will/should never be empty | 07:59:41 |
_hc | I guess I've forgotten how that code is working exaclty 🤣 | 08:02:00 |
_hc | reminds me of the experience I have from time to time: "wow, this code is whacked, who wrote this? Oh, me" | 08:02:41 |
glennmen | I changed it to use localesToUse parameter, this will contain an array of valid locales for this user or at least fallback english if available so it should never result in an empty Latest tab | 08:03:22 |
_hc | Bubu: do you think you'll be able to get rid of config['ndk_paths'] entirely? | 08:19:11 |
_hc | glennmen: I'll take a look at that MR again | 08:19:26 |
glennmen | thank you! | 08:21:25 |
_hc | Bubu: I've been thinking about reworking the checksums data storage to be more generic, and also be able to pull in other sources, like gradle-transparency-log, android-sdk-transparency-log, etc. | 08:47:07 |
_hc | seems tricky but doable. One question is whether there should be hardcoded defaults like we have now, then those can be extended/overridden by downloaded, then cached files | 08:48:08 |
Bubu | I have extracted the ndk checksums to fdroidserver/known_packages/ndk. Wrap a bit more logic around this and this could be auto-uodated. | 08:52:38 |
Bubu | In reply to@eighthave:matrix.org Bubu: do you think you'll be able to get rid of config['ndk_paths'] entirely? Should be relatively easy | 08:53:16 |
_hc | yay! | 08:54:12 |
_hc | that makes more of fdroidserver usable as an API | 08:54:40 |
_hc | you can see examples in the tests, basically to call varous methods, that config option has to be set | 08:55:11 |
Bubu | Weird. | 08:55:42 |
_hc | accepted_formats is probably the worst offender, that's going away soon | 08:56:27 |
_hc | glennmen: how about writing a test for your MR? Reading that code again makes me think it has to be tested. Seems like it might be easy, like just instantiating App() can calling setLocalized() with test data. there might even be related examples there, I forget | 08:59:06 |