In D25501#15033, @valerio.bozzolan wrote:

Maybe we can simplify things introducing a commodity "if viewer exists" method like:

This seems reasonable.

Noto seems like a reasonable choice. I personally really like DejaVu Sans a lot.

So apparently the ferret search engine doesn't really have any sort of dynamic ranking. The ranking is entirely based on how many ngrams match the query...with one singular exception - all user results are boosted above everything else.

I'll try to figure out a architecturally-appropriate way to do this. The users do get grayed out in the typeahead search, just not on the results pages.

can confirm, https://we.phorge.it/D25342?id=1111 does show an error for me in an incognito window.

I'm inclined to merge this. Any objections?

One concern that I have is that there isn't really a one-size fits all deployment. A single tiny VM image with the database and web server pre-installed would probably be ideal for evaluating and even for a tiny team's use. Once usage increases beyond a very basic level then users will fairly quickly outgrow a tiny VM and need at least a larger memory, more storage, possibly a separate database node, etc. Migrating from the tiny setup to a more robust setup might not be a fun process and I don't know how much we should encourage people to start out with a setup that won't scale very well for them.

I think there's an app-store style thing for hosted VMs in the big cloud providers

Glad you got it working and happy to help!

In T15322#15031, @aklapper wrote:

+1 on Content licensed under Creative Commons Attribution-ShareAlike 4.0 (CC-BY-SA) unless otherwise noted; code licensed under Apache 2.0 or other open source licenses.

In T15566#14934, @valerio.bozzolan wrote:

I've just found again this

https://github.com/cooperspencer/phorge/

In T15673#14931, @valerio.bozzolan wrote:

OH MY GOD FOR SOME REASONS THEY RE-OPENED OUR PROPOSAL \o/ IT'S HAPPENING!

https://pretalx.fosdem.org/fosdem-2024/talk/review/EXZNAUXQKMJ7HRYYRD7MKFXTTV8JFQDJ

Is this needed simply because PhabricatorApplicationSearchController doesn't actually have a try { } block to catch this particular case? Maybe we could improve the exception handling instead of return null?

In D25501#14443, @aklapper wrote:

This also needs rework because I still get a Query overheated when not being logged in and going to http://phorge.localhost/feed/transactions/query/all/

I think it might be applicable here and in several other places that follow the general pattern of “look up all subclasses of a given feature and then loop over all of them to run extension hooks “

The only trouble with maintaining a separate arcanist package is keeping the version in sync with phorge. Phabricator never followed anything like semantic versioning and there was never really any stable api version to depend on. Perhaps we could think about versioning and release cycles that follow a more predictable major version release cycle with API compatibility guarantees. It is certainly more convenient (for the upstream) to avoid all of that but it offloads quite a bit of burden on anyone trying to use the software.

TLDR; I think @aklapper is on the right track: I did some more digging and it looks like something similar to P24 is probably the only reasonably easy solution.

@aklapper you are now a member of Blessed Committers.

Since this was vetted by epriestley and previously by myself as well. I think that it exceeded the minimum standard for code review before it landed here.

Here's another one that I think deserves to be upstreamed: https://phabricator.wikimedia.org/T230787 adds context and search term highlights to fulltext search results. Currently Phorge and upstream Phabricator only show matching document titles with highlights on keywords in the title but not the body.
The efficiency of my solution is questionable, however, it's working well enough for Wikimedia's use of the feature.

Maybe this is a clue?

In T15653#13929, @valerio.bozzolan wrote:

Yeah so we are all for "let's just append as default and maybe think about something more for the future, if they ask"

What if it were possible to intentionally remove a tag by passing the tag prepended with - to signify removal? like ?tags=-ABCDE to remove a tag but otherwise the behavior would be to just append. I realize this doesn't have the same compatibility advantages but it does offer slightly more flexibility for use in ad-hoc workflows.

In D25066#8893, @valerio.bozzolan wrote:

Hi @20after4 :) Can I ask you if this could be useful in Wikimedia? I think yep

I like it but I haven't tested locally.

In T15096#11546, @avivey wrote:

Interesting.... Herald could probably start the flow, but there's still more parts needed (update vs create revision, extract summary and test plan)

I suspect there's also a problem of motivation: These things are lots of work, and the people who are capable of performing them - people who are comfortable with working with these tools - will not personally benefit from this stuff. The people who would benefit the most are people who are not involved in the project at all...

Cool extension! We still need to build a proper extension ecosystem for Phorge. Maybe we could start with a wiki page linking to known extensions, then you'll have a place to share your extension, should you decide to do so.

This is a feature that was invaluable on wikimedia's use of phabricator, however, proliferation of a lot of forms can be troublesome / difficult to manage. Ideally the forms system could use some improvements to make things more manageable.

More prior art: https://zuul-ci.org/docs/zuul/latest/gating.html#cross-project-testing

This is a nice contribution and definitely would be a valuable addition to the docs, IMO.

In T15364#8952, @avivey wrote:

"tooling for chains of diffs" might need its own topic, with a design of the what the whole thing should look like in an ideal world, and how to get to it.
Even if not all of the stuff fits nicely into Phorge, there's probably a lot that can work, and some conduit methods can be added to implement the rest.

Breaking this out from the discussion at T15364: arc land complaints summary by FreeBSD community

FWIW I think the test plan is configurable (that requirement can be disabled in phabricator's config server side which should remove it from commit messages in arc)

Relevant links:

I might be willing to take a stab at this but not sure when I'll have the time.

Phabricator's codebase is mostly high quality and fairly easy to jump into but I probably wouldn't recommend it for someone new to PHP :D

@ton welcome, glad to have your input here.

FWIW even in the case the milestones are strictly ordered, it's incredibly inconvenient to sort out a situation where you need to add one to the middle of the list - forcing the user to rename a bunch of them to get things straightened out.

In T15082#2028, @golyalpha wrote:

epriestley was very much against this idea but wikimedia's users loved it.

Do we have epristley's reasoning as to why he was against this? Might help in deciding about including this patch in Phorge.

In T15048#2234, @eax wrote:

IMHO if we are to do this it should be similar to "comment reactions" from github. Useful for e.g., +1 -1 but shouldn't generate visible transactions on the rest of the page.

To me 2 seems like the safest / easiest solution, even if it is a bit limited.