| Commit message (Collapse) | Author | Age | |
|---|---|---|---|
| * | Use non-blocking when using zlibHEADmaster | Christopher Baines | 37 hours |
| | | |||
| * | Use fibers for refreshing issue and branch data | Christopher Baines | 37 hours |
| | | |||
| * | Write metrics as a single string | Christopher Baines | 37 hours |
| | | | | | To reduce the number of system calls. | ||
| * | Queue writes since they seem to be delayed | Christopher Baines | 37 hours |
| | | |||
| * | Use cooperative scheduling for fibers | Christopher Baines | 37 hours |
| | | | | | To simplify things. | ||
| * | Retry canceling builds | Christopher Baines | 37 hours |
| | | | | | As sometimes this times out. | ||
| * | Increase timeout when canceling builds | Christopher Baines | 37 hours |
| | | |||
| * | Handle 404's while processing branch builds | Christopher Baines | 37 hours |
| | | |||
| * | Avoid issue with accessing the Git repo from fibers | Christopher Baines | 37 hours |
| | | |||
| * | Process the issue builds in parallel | Christopher Baines | 4 days |
| | | |||
| * | Add a way of controling the fibers for-each batch size | Christopher Baines | 4 days |
| | | |||
| * | Submit builds in parallel via fibers | Christopher Baines | 4 days |
| | | |||
| * | Submit branch builds via a fiber | Christopher Baines | 4 days |
| | | | | | Rather than a thread. | ||
| * | Add some fibers utils from the bffe | Christopher Baines | 4 days |
| | | |||
| * | Just use with-fibers-port-timeouts from the build coordinator | Christopher Baines | 4 days |
| | | |||
| * | Show the branch on the patches page | Christopher Baines | 4 days |
| | | |||
| * | Switch to Mumi/Debbugs issue titles | Christopher Baines | 4 days |
| | | | | | | Rather than using the data from Patchwork, this change also strips the [...] prefix. | ||
| * | Pre-compute the branch, and use the issue title as well | Christopher Baines | 4 days |
| | | | | | | This allows changing the branch that QA applies the patches to by changing the title of the issue. | ||
| * | Do less in the with-sqlite-cache transaction | Christopher Baines | 6 days |
| | | |||
| * | Simplify patches-view | Christopher Baines | 7 days |
| | | |||
| * | Increase timeout for submitting builds | Christopher Baines | 7 days |
| | | |||
| * | Fix non-master branch ordering | Christopher Baines | 10 days |
| | | |||
| * | Update the metrics approach | Christopher Baines | 2024-04-26 |
| | | |||
| * | Handle issues.guix GraphQL queries failing better | Christopher Baines | 2024-04-23 |
| | | |||
| * | Sort the patches before applying them to the git branch. | Vivien Kraus | 2024-04-16 |
| | | | | | | | | | | | | | | | | | | By default, the patches are not ordered. Because of the unreliability of e-mail delivery, they can be received completely out of order. Patchwork may also re-order them, and they may get ordered by commit author date (or, equivalently, e-mail Date header). Both possibilities lead to failure to apply patches if the order is incorrect for two patches that do not commute. The patch name contains a header: [bug#xxx,v12,i/n] where "i" is the variable part and indicates the patch number. Git pads "i" with zeros if needed. The string order of the names is thus the true commit order. * guix-qa-frontpage/manage-patch-branches.scm (create-branch-for-issue): Sort patches according to their "name". | ||
| * | Fix checking for unknown commits | Christopher Baines | 2024-04-02 |
| | | |||
| * | Fix ordering of simple-format call | Christopher Baines | 2024-04-02 |
| | | |||
| * | Add more logging around checking base branch changes | Christopher Baines | 2024-04-02 |
| | | |||
| * | Enable checking for base revision changes for all branches | Christopher Baines | 2024-04-02 |
| | | |||
| * | Fix not deleting some patch branches | Christopher Baines | 2024-04-01 |
| | | | | | When the patches were for a non-master branch. | ||
| * | Reduce the number of changes compared to master for branches | Christopher Baines | 2024-03-30 |
| | | | | | | As it was rather high, and it doesn't make sense to be merging a branch if there are lots of changes. | ||
| * | Try to better handle debbugs being unavailable | Christopher Baines | 2024-03-26 |
| | | |||
| * | Add a buffer to prevent deleting some branches that aren't found | Christopher Baines | 2024-03-17 |
| | | | | | To reduce instances of branches getting repeatedly deleted then recreated. | ||
| * | Make some module tweaks | Christopher Baines | 2024-03-17 |
| | | |||
| * | Ignore build-already-processed errors when canceling builds | Christopher Baines | 2024-02-26 |
| | | |||
| * | Account for builds to keep when canceling branch builds | Christopher Baines | 2024-02-22 |
| | | |||
| * | Add an information flow diagram | Christopher Baines | 2024-01-20 |
| | | |||
| * | Fix cancel-builds-not-for-revision | Christopher Baines | 2023-12-13 |
| | | | | | To actually cancel all the builds. | ||
| * | Cancel builds in a loop | Christopher Baines | 2023-12-13 |
| | | | | | | To ensure they're all canceled, since they might not all be returned initially when querying with #:relationship 'no-dependent-builds. | ||
| * | Make sure builds are canceled | Christopher Baines | 2023-12-13 |
| | | |||
| * | Cancel builds before submitting new ones | Christopher Baines | 2023-11-26 |
| | | | | | | This should handle the case where we want to cancel a build then submit a new one for the same output. | ||
| * | Cancel and resubmit builds where the derivation has changed | Christopher Baines | 2023-11-26 |
| | | | | | | | But the output remains the same. This will hopefully avoid some instances where the data service deletes the derivation as it corresponds to a removed revision and the build can't start because of the missing derivation. | ||
| * | Tweak error handling around talking to Mumi | Christopher Baines | 2023-11-06 |
| | | | | | | | I think requests to Mumi were being responded to with 502's, and that led QA to delete all the patch branches. This should hopefully stop that happening in the future. | ||
| * | Fix the package changes table on the branch page | Christopher Baines | 2023-11-03 |
| | | | | | And only display the "below systems" message if there are any. | ||
| * | Use a different symbol when waiting for build results | Christopher Baines | 2023-11-03 |
| | | | | | | And group these issues towards the top of the patches list since they potentially can be merged if there's not too many builds remaining. | ||
| * | Change text for the "waiting for build results" status | Christopher Baines | 2023-11-03 |
| | | |||
| * | Start to better describe when builds are waiting to happen | Christopher Baines | 2023-11-02 |
| | | | | | | Don't use the same styling as failed builds on the issue page, and also separate out the systems which don't count towards the issue status. | ||
| * | Add "i686-linux" back to systems to consider | Christopher Baines | 2023-11-02 |
| | | | | | As the substitute availability is better now. | ||
| * | Set higher rebuild number limit for patch series. | Julien Lepiller | 2023-11-02 |
| | | | | | | This allows building Java updates which impact 600 packages per architecture. | ||
| * | Use call-with-blocked-asyncs when calling call-with-zlib-input-port | Christopher Baines | 2023-11-01 |
| | | | | | Since it uses dynamic-wind that's incompatible with fibers. | ||