Conversation
6227309 to
8f9b5b3
Compare
| ### Tropical Geometry | ||
|
|
||
| ### Changes related to the package GAP | ||
|
|
||
| ### Changes related to the package Hecke | ||
|
|
||
| ### Changes related to the package Singular |
There was a problem hiding this comment.
why are there empty headers here without any content?
There was a problem hiding this comment.
This happens when a PR was tagged with a topic label, but not with a type label. The script finds non zero number of things under tropical geometry, so generates a heading for it. But it never finds a type label, so, neither the secondary heading, nor the PR itself is printed here. They end up in the TODO section. IMO, leaving it like this might be fine, so, we know where to (manually) insert something.
There was a problem hiding this comment.
I'd say it's better then to move these things with "missing labels into the right topic section, but in subsection #### TODO: missing type label or so
|
|
||
| - [#5267](https://github.com/oscar-system/Oscar.jl/pull/5267) Add `order_bound` keyword argument to `subgroup_classes` | ||
| - [#5284](https://github.com/oscar-system/Oscar.jl/pull/5284) Implement chamber counting algorithm for toric line bundles | ||
| - [#5288](https://github.com/oscar-system/Oscar.jl/pull/5288) Add `p_rump` for `GAPGroup` and extend `torsion_subgroup` for `GAPGroup` and `WeylGroup` |
There was a problem hiding this comment.
This has a "topic: groups" label, but is not sorted into the appropriate section above
There was a problem hiding this comment.
This should be fixed whenever #5279 is merged
Maybe it was unwise to add patches for this into that PR, from a keeping PRs on topic point of view, but its currently probably the fastest way of patching this.
|
Looking at some of the entries in the "TODO" categories, I realize that I don't know how many or what labels a pull request needs to make the script happy.
Is this correct? Is it written down somewhere? I so far I followed the two-label-approach from https://docs.oscar-system.org/stable/DeveloperDocumentation/changelog/ but apparently that's not enough anymore? |
As I noted in #5544 (comment), there is currently some issue with how these things are put into categories. I would assume that this also affects your examples |
8f9b5b3 to
06019bf
Compare
Your impression should be correct. Can you look at the new changelog ( 06019bf ), and if there is still unexpected behaviour, point out the exact PRs which are miscategorised ?
The two kinds of labels (topics, and types) have been expanded upon, but this is still in the dev docs, not yet in stable. https://docs.oscar-system.org/dev/DeveloperDocumentation/changelog/#Multi-level-topics |
|
|
||
| ### **TODO** insufficient labels for automatic classification | ||
|
|
||
| The following PRs only have a topic label assigned to them, not a PR type. Either assign a type label to them (e.g., `enhancement`), or manually move them to the general section of the topic section in the changelog. |
There was a problem hiding this comment.
Can you instead put the into their respective general section, but with a todo note each? I think it is easier to just manually remove todo notes than moving things around in this file.
I am asking because I see a bunch of things here, that don't fit into any subsection, e.g. updates of dependencies
There was a problem hiding this comment.
Day's event have started. I shall look at it again today at night.
Okay, sorry, I did not look at the most recent documentation. So the gist is that every pull request now needs three labels? And wouldn't it be good to have all the "type" labels start with "type:" and possibly all have the same colour? |
Yes that's how it has been since we introduced this system. (Well, you can get away with one label if it is
I don't want this for the colors: both "bug" and "enhancement" are types, but one is red and the other not, and I'd like to stay it that way. Wouldn't mind for them to share a common prefix, though. But perhaps we can not use "type" but rather something else, like "kind" or "category" -- the word "type" already is used heavily in our code and I'd like to avoid yet another overload of the meaning. Regarding the topics, note that they have either prefix |
2352b02 to
8c2a646
Compare
|
|
||
| Ping @HereAround , @friedemann2. | ||
|
|
||
| ~~Note: I can not test this version locally at the moment because with the latest master branch in Oscar.jl my julia is complaining about an incompatibility with GAP and I can somehow not resolve this issue with the usual updating procedures. I plan to clean this branch up a little further, once it works on my machine again.~~ Had to throw away my `~/.julia` folder; now it seems to work. Still working on the updates, though. | ||
|
|
There was a problem hiding this comment.
The release notes entry in the PR body needs to be a level two header:
Oscar.jl/dev/releases/release_notes.py
Line 179 in 91e2df4
but it is level one.
I will try changing the message and re-running it.
There was a problem hiding this comment.
After two more adjustments this did work:
- The itemize needs to be done with
- some textand not* some text - The list should not contain the PR link
#1234at the beginning since it will be added automatically.
Maybe the script can be adjusted further to be able to deal with these things but for now it did seem to work:
#### New features or extended functionality
- [#5032](https://github.com/oscar-system/Oscar.jl/pull/5032) Improve vector of minimal exponents method for cohomology computations
- [#5512](https://github.com/oscar-system/Oscar.jl/pull/5512) **Breaking:** Rename `all_cohomologies` to `sheaf_cohomology`
- [#5512](https://github.com/oscar-system/Oscar.jl/pull/5512) Support line bundle cohomology computation via local cohomology, including functorial aspects.
There was a problem hiding this comment.
I discussed this with @aaruni96 and he'll improve the script: instead of inserting "garbage", it will insert a suitable message about a parsing problem or so
I would suggest that we also make the script handle lists formatted with * instead of -.
6f6bf4d to
5819e27
Compare
|
@fingolfin @aaruni96 @lgoettgens What is the status of this now? If we want to create a release maybe tomorrow morning then this need to be completed, I just re-ran the script for 1.6.0 and it looks reasonable to me. |
|
I think the comments to improve the changelog script don't need to go in Oscar 1.6.0. Right now, with the current changes, the script produces a good changelog. The improvements suggested by @fingolfin above (to better handle strange cases, and accept However, I can most likely make those changes by EOD, in case those changes should be made ASAP. |
|
My comment was mostly regarding the changelog itself, not the script. I also think that further updates to the script can be done later. The current changelog in this PR looks good to me. |
5819e27 to
cd66b41
Compare
fingolfin
left a comment
There was a problem hiding this comment.
Looks good to me up to some very minor details in PR titles: I tweaked some (and saw someone else tweaked some more), so maybe we regenerate this a final time and then merge it?
cd66b41 to
1ec41c2
Compare
This comment was marked as outdated.
This comment was marked as outdated.
1ec41c2 to
e9496d6
Compare
lgoettgens
left a comment
There was a problem hiding this comment.
I just did two more minor consistency changes and regenerated. LGTM now.
Automated changes by create-pull-request GitHub action