Also want to add that we had several duplicate differential_drafts on our install as well. About 15 objectPHID-authorPHID combinations had duplicates, ranging from 2-10 duplicates each. Cleaning them up was pretty straightforward.

Perhaps I'm misunderstanding, but it seems like storage adjust could be made more efficient. Each time you alter a table, MySQL makes a copy of it, alters the copy then replaces the original with the modified one. If we perform several alters, this adds up to several table copies. On large tables this takes considerable time.

I've run into this as well occasionally at my work. We never isolated the conditions (seems to be some combination of windows and gitbash, but it works for some people)

Other installs should NOT run bin/storage adjust yet: we have some things that still need cleanup.

@g-p-g sorry for the delay on getting back to you on this.

In T1191#78120, @epriestley wrote:

Support Impact This has widespread impact, notably preventing import of repositories with an emoji anywhere in them with no workaround.

Thanks

Hmm. Our install is locked down anyways, you can only access it from our network, etc. I don't really care if dropped files are visible to everyone until this is resolved. If I removed these lines, do you foresee any universe destroying consequences?

Is there a short term solution? Writing documentation with images becomes very tedious if you have to fix the upload permissions on each file, and it's non immediately apparent if you made a mistake and one of the images is viewable only to you.

I think you mean un-Prototype :P

Crap. Since I also attached that file to this task, and you guys can view this task you can view the image.

Is there a way to view what objects a file is attached to?

Problem still occurs when updated to latest master.

Just kidding, got it to work on my test install

I actually can't seem to get it to reproduce at all. I can only figure that one of our developers restricted the permissions of the file he uploaded and then suffered amnesia about it.

Hmm, I might be wrong. Can anyone else see the image in the ponder question?

(9/25/2014 9:38:54 AM) epriestley: Note that we no longer plan to provide D9611 or the requested titular feature of T5024, per se, in the upstream.

Haha. I took a stab at the code to try and fix it. This was way easier than I thought it was going to be. Thanks!

@g-p-g Do you mind sending a screenshot of the data section (with points closed per day, etc.) on the one that goes negative?

Cleanup, more robust, and adding test

I was actually in the middle of submitting the exact same diff when you submitted yours :)

Can confirm this fixed the problem for me. Thanks!

The "%H" is being replaced with " H" (percent is turning to a space, which is mentioned in the comments around line 160.

I have also been seeing this.

T5474 should cover our use cases, this diff was something we did in the meantime so that we can do things when things move to certain columns until T5474 happens.

Tinkered with this again today, really frustrating trying to get phutil lint to work at all on windows.

Showing 173 orphaned commits, pruning them, then showing 0. The commits are now gone from git, but not from phabricator.

Fixed a bug thanks to cpa199. It should be ready to use. I'd appreciate feedback from anyone that uses it.

We implemented something like this for our install using custom fields that only show if a task has a certain project attached to it.

I did some git bisect-ing but ran into troubles.

Updating comment

Fix so that custom fields don't show up empty when editing

Okay just kidding, this is very bad.

This is something we use at our company, but it looks like no one else has asked for it. It's a useful way to extract files or folders as an archive out of a remote repository.

How do other applications do it? I copied this over from ManiphestCustomField because those were working for me.

Fixed by rP17badfacac311f5a8b9a2593525d6be620aeaa15

In case anyone cares:

I was bored, so I took a stab at this, using the files epriestley posted as a starting point. It's incredibly rough and ugly, but I think it builds the data in a way that mostly makes sense, though it is probably not very flexible and widely usable to other organizations. Sadly, I can't get the graph to show at all, and haven't worked out the "ideal points".

You sent an email to a bus? :P

I'd be willing take a stab at this if there is interest.

This worked wonderfully for me, thank you so much!

Got the join working, though had some fun where "column" can't be used, I think because it has special meaning in SQL. Also you were right about the Phid/PHID error.

• Fixing join, fixing columns not sticking due to "Phids" "PHIDs" typo, adding "Type a column or project name"
• Moving the "(on board Something)" into displayname, rather than the typeahead.

Couple questions:

It's not a symlink. It's a bash file.

Clean up

It seems this broke bin/phd. See T5388

Running unit tests and lint

Improving it so that checking local happens after checking for a global bin in path.

Ignore this one, I think that D9493 is a better solution.