Simplified judgement types

[deboer-tim]

Adds an optional simplified judgement type id to judgement types. These would be specified in cases where (e.g.) a team should not see the specific judgement that another team received.

The length of the property is unfortunate, but follows the naming pattern elsewhere.

** This is the first half of this change - judgements may need to be modified as well. We will not release a new version of the spec until we complete the second change or back this one out.

[johnbrvc]

An example would be nice in the Examples section that includes a judgement-type with a simplified judgement-type, and, the simplified judgement-type record.

[deboer-tim]

An example would be nice in the Examples section that includes a judgement-type with a simplified judgement-type, and, the simplified judgement-type record.

Added an example of TLE -> RE.

[deboer-tim]

I haven't understood the requirement why we need to know which judgements can be applied directly vs simplified ones, and why it wouldn't be simpler to use a different judgement type if that's required - i.e. having simplified_only seems like an unnecessary complexity to me. If others +1 I am fine with this though.

[niemela]

We discussed this. @nickygerritsen will write an update. If we get 2 approve (and no change requests) on it I will merge.

[nickygerritsen]

We discussed this. @nickygerritsen will write an update. If we get 2 approve (and no change requests) on it I will merge.

Seems I can't push to @deboer-tim s branch. I asked him to enable the option so I can since I have a change ready.

[deboer-tim]

That option was already on. I tried to figure out why you can't use it, and I'm guessing it's because you're listed as an outside collaborator and not a maintainer. According to google: "To promote an Outside Collaborator to a Maintainer on GitHub, you must first convert them to a member of the organization and then assign them a maintainer role"

Sure enough, you aren't even in the ICPC org - and I don't have authority to add you. @niemela ?

[niemela]

Sure enough, you aren't even in the ICPC org - and I don't have authority to add you. @niemela ?

I have invited @nickygerritsen to the org (and also @eldering for good measure). Is there someone else that is missing?

[nickygerritsen]

I think that didn't work like we wanted. I'm still getting this message:

You're not authorized to push to this branch. Visit https://docs.github.com/repositories/configuring-branches-and-merges-in-your-repository/managing-protected-branches/about-protected-branches for more information.

I think I need to become maintainer on this repo as well.

[niemela]

I think I need to become maintainer on this repo as well.

Now you are.

[nickygerritsen]

I think I need to become maintainer on this repo as well.

Now you are.

Thanks, the message went away. Pushing still fails...:

Pushing to github.com:deboer-tim/ccs-specs.git
error: Authentication error: Authentication required: You must have push access to verify locks
error: failed to push some refs to 'github.com:deboer-tim/ccs-specs.git'

I have no clue anymore, so here is my diff:

diff --git a/Contest_API.md b/Contest_API.md
index 7398e03..44be44c 100644
--- a/Contest_API.md
+++ b/Contest_API.md
@@ -913,7 +913,6 @@ Properties of judgement type objects:
 | penalty                         | boolean   | Whether this judgement causes penalty time. Required iff contest:penalty\_time is present.
 | solved                          | boolean   | Whether this judgement is considered correct.
 | simplified\_judgement\_type\_id | ID?       | Identifier of this type's simplified judgement type. When using simplified judgements, this is required iff simplified\_only is false.
-| simplified\_only                | boolean   | Whether this judgement is only used as simplified judgement type, never as normal judgement. Defaults to false.
 
 #### Known judgement types
 
@@ -972,17 +971,21 @@ For instance, in a contest where teams cannot see the specific reason another te
 might see their own judgement types, but judgements from other teams would return the corresponding simplified judgement
 type instead.
 
-If not using simplified judgements, the properties `simplified\_judgement\_type\_id` and `simplified\_only` must not be set.
+If not using simplified judgements, the property `simplified\_judgement\_type\_id` must not be set.
 
 A judgement type may be used both as original and simplified judgement type, but must then simplify to itself and have
 `simplified\_judgement\_type\_id` equal to `id`.
 For example, `AC` (aka correct) would typically map to `AC` also as simplified judgement type.
 
-The property `simplified\_only` must be set to `true` if a judgement type is only used as simplified judgement type.
-In that case `simplified\_judgement\_type\_id` must not be set. A typical example would be `RE` aka rejected.
-If `simplified\_only` is (per default) `false` then `simplified\_judgement\_type\_id` must have a value,
-that is, all original judgement types must simplify to something (which may be itself).
-Furthermore, the simplified judgement type must have the same `penalty` and `solved` values as the original jugdement type.
+If a system is interested in finding the set of judgement types that are only original judgement types, only simplified
+judgement types or both, one can use this logic:
+
+- The set of original judgement types are the ones that have `simplified\_judgement\_type\_id` set.
+- The set of simplified judgement types are the ones that appear in `simplified\_judgement\_type\_id`.
+- The set judgement types that are both is the intersection of these two sets.
+
+This assumes the system is using simplified judgement types. If it is not (i.e. if `simplified\_judgement\_type\_id` is not set
+for any judgement type), all judgement types are original only.
 
 #### Examples
 
@@ -997,8 +1000,7 @@ Returned data:
    "id": "RE",
    "name": "Rejected",
    "penalty": true,
-   "solved": false,
-   "simplified_only": true
+   "solved": false
 }, {
    "id": "TLE",
    "name": "Time Limit Exceeded",
@@ -1023,5 +1025,4 @@ Returned data:
    "penalty": false,
    "solved": true,
    "simplified_judgement_type_id": "AC"
-   "simplified_only": false
 }]

[johnbrvc]

I think I need to become maintainer on this repo as well.

Now you are.

Thanks, the message went away. Pushing still fails...:

Pushing to github.com:deboer-tim/ccs-specs.git
error: Authentication error: Authentication required: You must have push access to verify locks
error: failed to push some refs to 'github.com:deboer-tim/ccs-specs.git'

I think this is saying you need access to @deboer-tim 's REPO, not icpc. So, Tim has to grant you push access to HIS repo so you can update his fork that will get merged into the icpc ccs-specs.

[nickygerritsen]

I think I need to become maintainer on this repo as well.

Now you are.

Thanks, the message went away. Pushing still fails...:

Pushing to github.com:deboer-tim/ccs-specs.git
error: Authentication error: Authentication required: You must have push access to verify locks
error: failed to push some refs to 'github.com:deboer-tim/ccs-specs.git'

I think this is saying you need access to @deboer-tim 's REPO, not icpc. So, Tim has to grant you push access to HIS repo so you can update his fork that will get merged into the icpc ccs-specs.

But it does tell me this:

and normally in the DOMjudge repo at least I can push to other peoples repos.

[deboer-tim]

I sent an invite to the repo, just to see if that changes anything. 🤷🏼‍♂️

[nickygerritsen]

That DID work, I have pushed my change.

[johnbrvc]

That DID work, I have pushed my change.

By default, it seems logical. If "someone" forks a repo, what gives (even the maintainers) the right to push to that "someone's" fork? The "someone" has to grant that permission explicitly. Your DJ team must have granted each other permission to push to each other's forks.

[nickygerritsen]

That DID work, I have pushed my change.

By default, it seems logical. If "someone" forks a repo, what gives (even the maintainers) the right to push to that "someone's" fork? The "someone" has to grant that permission explicitly. Your DJ team must have granted each other permission to push to each other's forks.

Normally github has an option when you create a PR:

This option allows maintainers to push to your fork but only for the branch for that PR. It works like a charm normally, but apparently not in this repo.

[deboer-tim]

I like the changes 👍🏼, but simplified_only is still in the description at the top and example at the bottom. Want me to fix that, or do another commit?

In two places it says "If not using simplified judgements, the property simplified\_judgement\_type\_id must not be set." I'm tempted to say we should go stronger and say the property should be unused or even not exist in /access ... but I guess that's harder to implement and we don't do it elsewhere.

[nickygerritsen]

I like the changes 👍🏼, but simplified_only is still in the description at the top and example at the bottom. Want me to fix that, or do another commit?

Feel free to fix it.

In two places it says "If not using simplified judgements, the property simplified\_judgement\_type\_id must not be set." I'm tempted to say we should go stronger and say the property should be unused or even not exist in /access ... but I guess that's harder to implement and we don't do it elsewhere.

Yeah that's a bit harder I guess.

[deboer-tim]

Feel free to fix it.

Done.