18 Jan 2022 |
linsui | https://gitlab.com/fdroid/fdroiddata/-/merge_requests/10158 is merged just after your PR and since then there is no issuebot. | 07:37:44 |
Izzy | I first noticed that yesterday evening. I can skim the MR line for "last changes" to see when it was running last. | 07:38:12 |
Izzy | Last run I found was about a day ago (midnight 16th). 10h ago I found a bunch of commits added to some MR and issuebot did not kick in. | 07:42:59 |
Izzy | Didn't ever since. | 07:44:30 |
_hc | looking now | 09:03:57 |
_hc | Izzy: linsui turns out the error was easy to see: https://gitlab.com/fdroid/fdroiddata/-/jobs/1985257287 | 09:20:41 |
_hc | I'll push a fix MR now | 09:20:54 |
| obfusk joined the room. | 09:29:28 |
Izzy | Thanks! | 10:11:27 |
Izzy | Might be easy to see. The difficulty, as usually, is knowing where to look #D | 10:12:02 |
_hc | the "pages" job on fdroid/fdroiddata is issuebot. It can't be called "issuebot" because GitLab CI jobs have to be called "pages" in order to publish to GitLab Pages... | 10:22:52 |
_hc | hi obfusk good to see you around here | 10:25:44 |
grote | In reply to @grote:matrix.org Vendoring this as a quick fix to unbreak CI is fine with me. jcenter still down :( | 11:19:28 |
grote | About the indexV2 version code discussion: Can't we just leave the version code and add a suffix in case there's more than one? | 11:35:32 |
jochensp | I think the question is how fdroidserver determines the suffix (or any version) in case of a binary repo | 11:46:22 |
_hc | Izzy: linsui issuebot is back https://gitlab.com/fdroid/fdroiddata/-/merge_requests/10041#note_813537645 | 11:47:10 |
_hc | In reply to @grote:matrix.org About the indexV2 version code discussion: Can't we just leave the version code and add a suffix in case there's more than one? you mean a suffix just for the key in {packages: {versions: {"versionCode-suffix": {}}}} ? | 11:48:32 |
grote | In reply to @_oftc_jochensp:matrix.org I think the question is how fdroidserver determines the suffix (or any version) in case of a binary repo isn't it hecking the version codes in the APKs it finds? If so, it could add a suffix as soon as it finds a second one. | 11:49:08 |
grote | In reply to @_oftc_jochensp:matrix.org I think the question is how fdroidserver determines the suffix (or any version) in case of a binary repo * isn't it checking the version codes in the APKs it finds? If so, it could add a suffix as soon as it finds a second one. | 11:49:16 |
grote | We have an audio meeting in ~1h about the index, right? | 11:49:31 |
jochensp | meeting: yes | 11:49:45 |
_hc | is that key only used for the diff format? seems like we should just use something else unique that has meaning, like filename or sha256 | 11:49:58 |
jochensp | no, the key is meant to get an order in the list | 11:50:54 |
jochensp | reducing the diff is only minor | 11:51:19 |
_hc | ah I see. | 11:51:40 |
_hc | if the diff part is minor, then why not just make versions: a list? | 11:52:07 |
_hc | I think the diff won't be minor for a large archive like https://f-droid.org/archive | 11:52:43 |
_hc | where versions: will have like 30 or 50 entries | 11:53:02 |
jochensp | maybe, I was optimizing for the repo for now | 11:53:06 |
grote | yeah it might make sense to use a dict and keep the keys short, if they don't get parsed. | 11:53:25 |