Ah, sorry @20after4 if I broke "your" Wikimedia Phabricator ihih

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"

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.

Hi @Matthew can I help in landing this useful change?

You have green light, feel free to land and thanks

@valerio.bozzolan: I've lost track a bit given the latest comments. Shall I commit the accepted version, or are changes required?

Where to find the test plan? (And is expanding a hard requirement to get this reviewed?)

Maybe we can expand the test plan:

Thanks for reporting.

After applying this patch locally, after selecting Import .ics file, I can confirm that /calendar/import/edit/ does not show an URI field anymore, and that selecting an ICS File and importing it successfully creates calendar/import/8/ with expected name, date, and time.

Hi @bekay, feel free to look at D25408. It should fix your original problem. Thanks for your reporting

I've probably understood what is going on here

What about something like this?

Hoping to be useful, let's land this.

move comment

Yeah, the first run of arc lint creates the file, and subsequent runs do not update it.

smallcow test case added.

add useful comment (at least in my opinion) about $idx

I'm trying to propose two unit tests, but I'm not able to make them "fail" before the change. The unit always says "OK" so I think I'm not useful :( But here as reference:

Replace $m with $matches, as per review.

git rebase master

I triage this to Wishlist since it's indeed useful to some people but it's somehow still controversial. In order to attract more eyes, adding also Discussion Needed.

Feel free to land \o/

What about this small change? So, the checkbox has not a related "left" label, and has a more descriptive "right" label.

Tested before/after, thanks :)

It's a Git repository branch called 201709 (see downstream) but still no idea how to trigger the exception.
If anyone has an idea what to do to reach code in DiffusionRequest::generateURI() or DiffusionCommitBranchesController, please share.

I've taken the habit of updating to "latest RC" whenever I make an RC cut (every 3-4 months?), and then leaving it there until I feel like updating to master.
"I feel like updating to master" ususally means "there's something interesting there", and unless it's only interesting to this install, I tend to also cut the RC while I'm at it (or wait for when I do that).

Just for reference, see related T15449 and Q56.

Or, additional approach could be to have a small periodic script to watch for change on diffusion.branchquery Conduit call and than to trigger a build plan through Conduit.

looks good

great - see D25446

@avivey Yes this seems to work! Thank you

@fgaz can you try this and see if it it works for you?

I'm looking at PhutilLibraryMapBuilder:analyzeLibrary(), and it kinda looks like:

• if $this->fileSymbolMap is null, we call analyzeLibrary()`
• We load the data from the file if it's there
• We update the data
• We store it in $this->fileSymbolMap
• and we always save it to disk

There's no return in analyzeLibrary`, so if we got there, we'll save to disk. And there's no other write to fileSymbolMap, so we'll always get there.

The command does create the file, but arc lint tries to modify it anyway: