Summary:
We have `max_allowed_packet` 1 GiB but our replication dies if the query is longer than unknown value (it dies with 293 MB long query).
Anyway, there's no reason why we should not save the cache if you have small `max_allowed_packet`.
Test Plan: Lowered `$size` to 100, deleted cache from DB, displayed changeset, verified issued queries in DarkConsole, verified DB.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3390
Summary:
I like systems that just work. It is possible to store files larger than max_allowed_packet in MySQL and we shouldn't demand it.
It also fixes a problem when file was smaller than `storage.mysql-engine.max-size` but its escaped version was larger than `max_allowed_packet`.
Test Plan: Reduced the size to 5e4, uploaded 90 kB file, checked the queries in DarkConsole, downloaded the file.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3392
Summary: We recently opted for 'security.alternate-file-domain' and we have some hotlinks to the original domain.
Test Plan: Enabled 'security.alternate-file-domain', observed redirect.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3380
Summary:
This does a few things:
- Allows you to flag pastes. This is straightforward.
- Allows Applications to register event listeners.
- Makes object action lists emit a 'didrenderactions' event, so other applications can add more actions. The Flags application injects its action in this way. This should generally make it much easier to add actions to objects when we add new applications, with less code duplication and better modularity. We have a really hacky version of this in Differential that I want to get rid of in lieu of this more general approach. I'm going to make object lists do the same thing, so any application can jump in and add stuff.
Test Plan: Flagged and unflagged pastes. Viewed home page, differential, flags list.
Reviewers: vrana, btrahan
Reviewed By: btrahan
CC: aran
Differential Revision: https://secure.phabricator.com/D3377
Summary: Permits the name and langauge of a paste to be edited. This will eventually allow the visibility policy to be edited as well.
Test Plan: Edited name/langauge of some pastes. Tried to edit a paste I didn't own, was harshly rebuffed.
Reviewers: vrana, btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T1690
Differential Revision: https://secure.phabricator.com/D3376
Summary:
We have this hybrid "create / last few pastes" landing screen right now but I ~never use the list at the bottom and it makes the controller kind of complicated. I want to let you edit pastes too, and this generally simplifies things.
Also makes the textarea monospaced and cleans up the fork logic a bit.
Test Plan: Created, forked pastes. Viewed paste lists. Viewed pastes.
Reviewers: vrana, btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T1690
Differential Revision: https://secure.phabricator.com/D3375
Summary:
When logging in as an LDAP user for the first time (thus registering), a DAO exception was being thrown because PhabricatorLDAPRegistrationController wasn't passing in a username to PhabricatorUser::setUsername().
Somewhat separately, since either the PHP LDAP extension's underlying library or Active Directory are returning attributes with lowercased key names, I have to search on sAMAccountName and look for the key samaccountname in the results; this is fine since the config allows these to be defined separately. However I found that PhabricatorLDAPProvider::retrieveUserName() was attempting to use the search attribute rather than the username attribute. This resolves.
Test Plan: Tested registration and login against our internal AD infrastructure; worked perfectly. Need help from someone with access to a functional non-AD LDAP implementation; I've added the original author and CCs from D2722 in case they can help test in this regard.
Reviewers: epriestley, voldern
Reviewed By: voldern
CC: voldern, aran, Korvin, auduny, svemir
Differential Revision: https://secure.phabricator.com/D3340
Summary:
See T1602#15.
I don't intend to land this right away, because I'm not sure I won't
need another change or two to the rendering code. Keeping it here so I
won't forget about it.
Test Plan: iiam
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Maniphest Tasks: T1602
Differential Revision: https://secure.phabricator.com/D3240
Summary: See comment inline. We should fix this properly but it goes faily deep so I just stopped the bleeding for now. It's OK if we end up with a silly-looking file tree view for bizarre edge cases.
Test Plan: Created a problem diff as described, verified it no longer threw.
Reviewers: vrana, btrahan
Reviewed By: vrana
CC: aran, Avish
Maniphest Tasks: T1702
Differential Revision: https://secure.phabricator.com/D3373
Summary: If the commit has a known author but that author is different from the revision author or the revision doesn't exist, we'll throw away the commit author and then raise an audit for "Owners Not Involved". Instead, we should just use the commit author in all cases.
Test Plan:
Debugged this with Zor in IRC, he reported it fixed his issue.
Before:
http://pastie.org/4574680
("Author" is empty.)
After:
http://pastie.org/4574712
("Author" is correctly recognized.)
Reviewers: jungejason, nh, vrana, btrahan
Reviewed By: vrana
CC: aran
Differential Revision: https://secure.phabricator.com/D3374
Summary:
I quite often wonder what's inside these gaps displayed in changeset. In which method I am? Is it a while loop or a foreach? Maybe I'm in a class but the project doesn't have a sctrict policy of one class per file so what's the name of the class?
I've experimented with bunch of rules:
- Always display 0 indentation: useless for one class per file.
- Always display 1 indentation: weird inside global functions.
- Display closest lower indentation: works best.
I'm not sure about highlighting the context. I like highlighting but maybe in this case subtler monochrome text will work better.
Test Plan: Browsed bunch of diffs, loved it.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3371
array_diff(): Argument #1 is not an array at [/var/www/phabricator/phabricator/src/applications/feed/PhabricatorFeedStoryPublisher.php:109]
Auditors: alanh
Summary:
Actions you made no longer show up in the lighting-bolt
dropdown. I didn't touch realtime notifications but they're transient
enough that it shouldn't matter too much?
I wonder, though, whether it would be more useful to have the
notifications still present but automatically marked read.
Test Plan:
Create notifications; muck around in the database; check that
the dropdown and list pages are displaying correctly.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Maniphest Tasks: T1402
Differential Revision: https://secure.phabricator.com/D3360
Summary:
Just a bunch of copy-pasta from D2884. I suppose this calls for
a refactoring at some point...
Test Plan:
Make a bunch of updates, some from different users; check
notifications dropdown and list.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3361
Summary: This has been deprecated for quite a while and I'm pretty sure there are no callsites in the wild since this tool doesn't get much use outside of Facebook.
Test Plan: grep
Reviewers: vrana, btrahan, meitros
Reviewed By: vrana
CC: aran
Maniphest Tasks: T603
Differential Revision: https://secure.phabricator.com/D3195
Summary:
Adds a flexible navigation menu to diffs that shows you your current position in the diff.
Anticipating some "this is the best thing ever" and some "this is the wosrt thing ever" on this, but let's see how much pushback we get? It seems pretty good to me.
Test Plan: Will attach screenshots.
Reviewers: vrana, btrahan
Reviewed By: vrana
CC: aran
Maniphest Tasks: T1633, T1591
Differential Revision: https://secure.phabricator.com/D3355
Summary:
We have some issues with Elastic search (or maybe it's SMC) causing that indexing sporadically doesn't work.
Throwing in indexing stops the workflow and is annoying.
Not indexing doesn't have fatal consequences for the user and we can (and probably should) postpone it.
Test Plan: Thrown, looked at log.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3350
Summary: I had no idea what checkered is.
Test Plan: Flagged revision, flagged task.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3346
Summary:
We have /differential/filter/drafts/ but nobody knows about it.
This diff displays the draft only if there is no flag to not waste space.
Test Plan: /differential/filter/revisions/
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin, alanh
Differential Revision: https://secure.phabricator.com/D3324
Summary:
Rendering method shouldn't load data.
The view probably shouldn't load data either because it is a job for component (object that both loads data and displays them) but we don't have that concept in Phabricator.
This at least improves the architecture a little bit.
Test Plan: /differential/
Reviewers: epriestley
Reviewed By: epriestley
CC: alanh, aran, Korvin
Differential Revision: https://secure.phabricator.com/D3325
Summary:
This is another experiment for reducing reviewers response time.
I stole the idea (and colors) from [[ http://www.reviewboard.org/media/screenshots/2009/02/02/dashboard.png | ReviewBoard ]].
I actually quite like it (except when everything is red) and I can image that people will review just to have better color balance.
The code is not production ready for these reasons:
- We load holidays again and again for each revision. I couldn't cache it to static variable because it could persist multiple requests, right?
- I don't know how to expand height to the whole cell (I'm really bad in CSS).
- CSS rules are probably in wrong file.
- We probably want to use different colors.
This is how it looks:
{F16406}
Test Plan: Displayed revision list.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3190
Summary:
- Fix width, corresponding to wider sprites.
- Sprite the "Audit" icon.
- Mark the meta-application as device-ready.
- Fix some collapse/expand bugs with the draggable local navs.
- Add texture to local nav.
- Darken the application nav to make it more cohesive with the main nav. I think this is an improvement?
Test Plan: See screenshots.
Reviewers: chad, btrahan
Reviewed By: btrahan
CC: aran, netfoxcity
Maniphest Tasks: T1569
Differential Revision: https://secure.phabricator.com/D3338
Summary: See T1677. I think wanting bots to be able to post comments without sending email is a pretty reasonable use case. Eventually we should probably support this more broadly and maybe protect it with permissions (normal users maybe shouldn't be able to do this?) but we can wait for use cases.
Test Plan: Made comments with and without "silent". Verified that the non-silent comment sent email, and the silent comment did not.
Reviewers: btrahan, vrana
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T1677
Differential Revision: https://secure.phabricator.com/D3341
Summary: Currently, if no placeholder is configured we always move "Cc" up to "To", even if we have a valid "To". Instead, move "Cc" to "To" only if there's no "To" and no placeholder.
Test Plan: Sent email with "to" and "cc", email with "cc" only with a placeholder, and email with "cc" only without a placeholder. Verified recipients ended up in the right location in all cases.
Reviewers: nh, btrahan
Reviewed By: btrahan
CC: klimek, aran
Differential Revision: https://secure.phabricator.com/D3342
Summary:
The logic here was swapped - new file should be on the right side.
Plus we had a fatal for VS = -1 where new file should be on left.
Test Plan:
Downloaded raw diff of:
- base VS change
- change VS change
- change VS change with unmodified file
Reviewers: epriestley, btrahan
Reviewed By: btrahan
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3333
Summary: We need to use commit diff in some links and manual diff in some others.
Test Plan: Displayed revision, verified that the action link uses commit diff.
Reviewers: epriestley, nh
Reviewed By: nh
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3300
Summary: See T1665. If you have a directory named 'readme', we try to read it as a README.
Test Plan: Created a directory named 'readme', hit a similar fatal to the one in T1665, applied this patch, everything worked great.
Reviewers: btrahan, vrana
Reviewed By: vrana
CC: aran
Maniphest Tasks: T1665
Differential Revision: https://secure.phabricator.com/D3312
Summary:
If I use my own selector then it doesn't respect Phabricator config.
Also I hated this method.
Test Plan: Used default selector, displayed revision.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3307
Summary:
It's kind of nice to type `s explode` in jump nav and have it
just work. This involves weakening a bunch of the request parameter
checks, but the only real extra assumption is that language defaults to
PHP... which is not that big of a stretch, and it's not like we know
about any other languages' documentation.
It'd be better to have builtins be more first-class and less awkward
hack, but that seems hard.
Test Plan: Search for symbols.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D3302
Summary:
The "Running Daemons" tab in `/daemon` links to
`/daemon/log?show=running/`, which doesn't work. No other side nav tries
to link to a URL with a query string, and I don't know if this ever
worked. This fixes that in a minimally invasive way.
NOTE: Daemons run when a good man goes to war.
Test Plan: View `/daemon` and click on tabs.
Reviewers: epriestley, nh
Reviewed By: nh
CC: aran, Korvin, nh
Differential Revision: https://secure.phabricator.com/D3301
Summary:
This is the first time I've ever had CSS actually work like it promises it does (i.e., markup the "right" way and then you don't have to change the markup later).
Since I laboriously laid this whole thing out with <divs> originally, I was able to just override some of the styles and make the layout reasonable for devices.
The only differences for existing forms are:
- No colon after labels (looks cleaner anyway).
- Non-error required text is no longer a red star but a the grey word "Required" (this is clearer).
Test Plan:
Viewed paste form on a phone.
Viewed ~20 other forms on the site to verify that I didn't break anything.
Reviewers: btrahan, chad, vrana
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T1569
Differential Revision: https://secure.phabricator.com/D3298
Summary:
I just put them in the property table instead of a list at the foot, they looked weird down there and were too bulky relative to their importance.
This won't scale great if someone forks a paste ten thousand times or whatever, but we can deal with that when we get there.
Also clean up a few things and tweak some styles,
Test Plan: Looked at forked, unforked pastes.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Differential Revision: https://secure.phabricator.com/D3295
Summary:
- Add a PhabricatorApplication.
- Make most of the views work well on tablets / phones. The actual "Create" form doesn't, but everything else is good -- need to make device-friendly form layouts before I can do the form.
Test Plan: Will attach screenshots.
Reviewers: btrahan, chad, vrana
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T1569
Differential Revision: https://secure.phabricator.com/D3293
Summary:
This case is unusually complicated because there are more rules than most objects will have.
- Edits are either "joins", "leaves" or "other edits".
- "Joins" require "can join" or "can edit".
- "Leaves" don't require any policy.
- "Other edits" require "can edit".
- You can't edit away your ability to edit.
- You //can// leave a project that you wouldn't be able to rejoin.
Things I'm going to add:
- Global log of policy changes.
- `bin/policy` script for undoing policy changes.
- Test coverage for these rules.
Test Plan: Made various project visibility edits with various users, joined / left projects, etc. I'll add more complete coverage in the next diff.
Reviewers: btrahan, vrana
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T603
Differential Revision: https://secure.phabricator.com/D3270
Summary: See D3291.
Test Plan: Ran `phpast.getast` via API.
Reviewers: alanh
Reviewed By: alanh
CC: aran
Differential Revision: https://secure.phabricator.com/D3292
Summary:
You can now embed countdowns in Remarkup! Not sure what it's
useful for, but there you have it.
Also I may have made a hash of the markup code; I don't really know what
I'm doing.
Test Plan: Make a new countdown, put `{C###}` in a Differential comment.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Maniphest Tasks: T1053
Differential Revision: https://secure.phabricator.com/D3290
Summary:
Create `phpast.{version,getast}` methods for calling xhpast
with `--version` and with source code as input, respectively.
Test Plan:
Run `arc call-conduit` a bunch of times; delete xhpast; run
it a bunch more times.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Maniphest Tasks: T1534
Differential Revision: https://secure.phabricator.com/D3289
