Perhaps ConduitAPI_project_query_Method should raise an error if a specified slug doesn't exist?

While I think D9619 is good, I don't think this works. Particularly, if I type --project FOO, but the project really has slug #foo, the query will normalize the slug and find the project, but then this code will incorrectly determine that the slug wasn't found because the case is different. Is that accurate?

Elsewhere, we return a separate map (for example, the identifierMap on diffusion.querycommits). We could do this here, but it would break any existing callsites of projects.query since the top-level result format would change structure. Not the end of the world (we probably need to do this eventually to add pagination information) but seems like a lot of breaks for arc todo.

I don't think projects.query should fail on invalid slugs, either, since it's inconsistent with other methods and less useful than a method which returns partial results.

We could also return matchingSlugs on the results, but I don't particularly like the idea that the "object dictionary" we hand you had some non-object / query-dependent data. I don't think this is terrible, but it feels a bit surprising and is inconsistent with diffusion.querycommits.

In D9457#18, @epriestley wrote:

While I think D9619 is good, I don't think this works. Particularly, if I type --project FOO, but the project really has slug #foo, the query will normalize the slug and find the project, but then this code will incorrectly determine that the slug wasn't found because the case is different. Is that accurate?

Are slugs always case-insensitive? We could pass them through strtolower?

I think we should treat the slug rules as a black box -- they're relatively complicated:

https://secure.phabricator.com/diffusion/P/browse/master/src/infrastructure/util/PhabricatorSlug.php;f52fbf61170d5a64fef9f16f11a255b1f3cdb0aa$5-32

I don't quite understand the issue. The query doesn't seem to normalize the slugs.

> echo '{"slugs": ["foo"]}' | arc --conduit-uri='http://phabricator.joshuaspence.com' call-conduit project.query
Waiting for JSON parameters on stdin...
{"error":null,"errorMessage":null,"response":{"PHID-PROJ-ttomlhslujpx5sdpbu2c":{"id":"1","phid":"PHID-PROJ-ttomlhslujpx5sdpbu2c","name":"Foo","members":["PHID-USER-cb5af6p4oepy5tlgqypi"],"slugs":{"1":"foo","2":"bar"},"dateCreated":"1402422720","dateModified":"1402422728"}}}

> echo '{"slugs": ["FOO"]}' | arc --conduit-uri='http://phabricator.joshuaspence.com' call-conduit project.query
Waiting for JSON parameters on stdin...
{"error":null,"errorMessage":null,"response":[]}

Oh -- well, it should be normalizing -- that is, --project Diffusion, --project 'Quality Assurance', etc., should work.

One approach would be to introduce project.normalizeslugs or something?

In D9457#22, @epriestley wrote:

Oh -- well, it should be normalizing -- that is, --project Diffusion, --project 'Quality Assurance', etc., should work.

One approach would be to introduce project.normalizeslugs or something?

That means we'd need to do two Conduit calls... first to normalize the slugs and then to query the project. What about introducing project.queryslugs instead?

In D9457#23, @joshuaspence wrote:

In D9457#22, @epriestley wrote:

Oh -- well, it should be normalizing -- that is, --project Diffusion, --project 'Quality Assurance', etc., should work.

One approach would be to introduce project.normalizeslugs or something?

That means we'd need to do two Conduit calls... first to normalize the slugs and then to query the project. What about introducing project.queryslugs instead?

@epriestley, how do you think that I should tackle this problem?

We don't actually call project.query anywhere, so let's just modernize it to work like the most modern methods (e.g., diffusion.querycommits), and return:

• a list of results in data (the current results of the entire call);
• paging parameters; and
• a slugMap mapping any queried slugs to their associated projects PHIDs.

We need to do this sooner or later anyway.

Just pinging you on this which should be okay (I think) after D9619.

I think this should use slugMap so it continues working after we start normalizing slugs?