Summary:
Ref T4705 (there are also some other adjacent related tasks dealing with URIs).
Currently, we issue a "get repositories matching URIs: ..." query by loading every possible repository and then checking their URIs in PHP.
Instead, put URIs in a separate table. I plan for each repository to potentially have multiple URIs soon, so this prepares for that.
Test Plan:
- Ran migrations.
- Looked at index table, made sure it appeared sensible.
- Ran some queries by `uri` to find repositories, found the repositories I expected.
- Updated the remote URI of a repository, saw queries / index update appropriately.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4705
Differential Revision: https://secure.phabricator.com/D15005
Summary: Fixes T10096. The `**` was being parsed as part of the hashtag, so `**#asdf**` interpreted `#asdf**` as a hashtag.
Test Plan: Unit test; bolded stuff with hashy contents.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10096
Differential Revision: https://secure.phabricator.com/D15006
Summary:
Fixes T10135. When the viewer is a member of no projects, specify the constraint type as a new "EMPTY" type.
When a query has an "EMPTY" constraint, fail fast with no results.
Test Plan:
- Viewed a viewerprojects() query result set as a user in no projects.
- Before patch: got a lot of hits. After patch: no hits.
- Viewed a normal result set, no changes.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10135
Differential Revision: https://secure.phabricator.com/D15003
Summary: A lot of confusing stuff happens when you create a milestone or subproject which isn't explained clearly in the UI. This is causing more harm than good on the balance since we're still figuring out how to move forward here. Just turn it off for now until we're closer to pushing it forward.
Test Plan: Viewed a project, no more UI for subprojects or milestones.
Reviewers: chad
Reviewed By: chad
Differential Revision: https://secure.phabricator.com/D14999
Summary:
See IRC. We're supposed to repair configuration, but if custom validators throw a generic `Exception` or use `PhutilTypeSpec` to do a check, we may explode way harder than we intend to.
Instead, soften these exceptions into validation exceptions so we repair configuration, raise a setup issue, and continue.
Test Plan: {F1059609}
Reviewers: chad
Reviewed By: chad
Differential Revision: https://secure.phabricator.com/D14998
Summary:
Ref T10121. This doesn't work at all at HEAD, and even when it did it was mostly just confusing to installs with unexpected setups where Phabricator is receiving mail at `@mycompany.com` and this is colliding with real addresses.
It might make sense to restore it some day after the next Conphernece update, but just strip it out for now. Since it doesn't work anyway, I'm pretty confident no one is using it.
Test Plan:
- Before patch: send mail to `dog@local.phacility.com`, got a policy error from Conpherece.
- After patch: sent mail to `dog@local.phacility.com`, got a correct "no routable recipients" error.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10121
Differential Revision: https://secure.phabricator.com/D14997
Summary:
Ref T10121. If a user sends mail with duplicate headers, like:
```
X-Duplicate: A
X-Duplicate: B
```
...and we process it with `mail_handler.php`, we may end up with `array('A', 'B')` as the header value. When we try to write this back into an error response mail, it fails.
Test Plan:
- Generated a message with duplicate headers.
- Piped it into `mail_handler.php` with `--process-duplicates` and `--trace` to get a look at it.
- Faked an exception.
- Before patch: bad error email.
- After patch: clean error email showing multiple header values.
```
$ ./bin/mail show-outbound --id 12386
PROPERTIES
ID: 12386
Status: queued
Related PHID:
Message:
PARAMETERS
sensitive: 1
is-error: 1
force: 1
subject: Error Processing Mail (Exception)
raw-to: ["epriestley@phacility.com"]
HEADERS
X-Phabricator-Sent-This-Message: Yes
X-Mail-Transport-Agent: MetaMTA
X-Auto-Response-Suppress: All
TEXT BODY
Your email to Phabricator was not processed, because an error occurred while
trying to handle it:
Exception: TEST
-- Original Message Body -----------------------------------------------------
testy testy
-- Original Message Headers --------------------------------------------------
from: Evan Priestley <epriestley@phacility.com>
content-type: text/plain; charset=us-ascii
content-transfer-encoding: 7bit
x-smtp-server: smtp.gmail.com:epriestley@phacility.com
subject: test outbound mail
message-id: 7isvptmllqvdvtdxthvdwzg3woj5au7csyuh3hopypjv6y6hqb32qm4bcrd4jtid
x-universally-unique-identifier: 4E489E20-F674-49B2-94BA-0DE44F504EAA
date: Mon, 11 Jan 2016 09:50:12 -0800
date: Mon, 11 Jan 2016 09:50:13 -0800
date: Mon, 11 Jan 2016 09:50:14 -0800
date: Mon, 11 Jan 2016 09:50:15 -0800
to: epriestley@yghe.net
mime-version: 1.0 (Mac OS X Mail 8.2 \(2104\))
HTML BODY
(This message has no HTML body.)
```
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10121
Differential Revision: https://secure.phabricator.com/D14996
Summary: Fixes T8826. Git tracks an "author date", which may be different from the "committed date". We don't currently extract/show this; do so.
Test Plan: {F1059235}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T8826
Differential Revision: https://secure.phabricator.com/D14995
Summary:
Fixes T10117.
- I accidentally broke setting `null` to unassign tasks at some point when I added richer validation.
- Raise a better error if the user passes junk.
Test Plan:
- Unassigned a task via API and web UI.
- Reassigned a task via API and web UI.
- Tried to do an invalid assign via API, got a sensible error.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10117
Differential Revision: https://secure.phabricator.com/D14992
Summary: Ref T4245. Full commit display names (like `rPaaaa`) are going to be obnoxious soon in some cases (e.g., `rPaaaa` becomes `R123:aaaa`, which is much uglier) so reduce how often we show the repository in cases where it isn't really necessary to include it.
Test Plan:
- Saw no more `rX` on repository list view for Git/Mercurial (still present for Subversion).
- Saw no more `rX` on various repository detail views, except when referencing other commits (e.g., mentions).
- Grepped for removed `getShortName()`.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4245
Differential Revision: https://secure.phabricator.com/D14990
Summary:
Ref T4245.
- Rename "Clone/Checkout As" to "Short Name" in the UI.
- Allow any repository to have a short name, not just hosted repositories.
Test Plan:
- Ran migration.
- Reviewed old transactions, saw they looked good.
- Edited an existing repository's short name.
- Gave an imported repository a new short name.
- Removed a repository's short name.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4245
Differential Revision: https://secure.phabricator.com/D14989
Summary:
Ref T4245. This allows `bin/repository update bread` to work, in addition to `rBREAD`, `R123`, `123`, `BREAD`, etc., if a repository has a short name set.
This primarily affects CLI commands (like `bin/repository`) and Conduit API calls. It has no normal user-facing impact.
Test Plan: Ran `bin/repository update bread` and such.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4245
Differential Revision: https://secure.phabricator.com/D14988
Summary: Fixes T10116.
Test Plan:
- Prior to patch: Control/Command + Enter submitted form but ignored actions.
- After patch: verifid command + enter submits actions.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10116
Differential Revision: https://secure.phabricator.com/D14991
Summary:
Via HackerOne. The use of `$key` here should be `$extended_key`.
Exploiting this requires a very unusual group of objects to be subjected to extended policy checks. I believe there is no way to actually get anything bad through the policy filter today, but this could have been an issue in the future.
Test Plan:
- Added a unit test which snuck something through the policy filter.
- Fixed use of `$extended_key`.
- Test now passes.
Reviewers: chad
Reviewed By: chad
Differential Revision: https://secure.phabricator.com/D14993
Summary:
Fixes T7938.
- Primarily, users can currently shoot themselves in the foot by putting `../../etc/passwd` and other similar nonsense in these fields (this is not dangerous, but also does not work). Require sensible names.
- Enforce uniqueness so these names can be used in URIs and as identifiers in the future.
- (This doesn't start actually using them for anything fancy yet.)
Test Plan:
- Gave several repositories clone names: a valid name, two duplicate names, an invalid, name, some with no names.
- Ran migrations.
- Got clean conversion for valid names, appropriate errors for invalid/duplicate names.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T7938
Differential Revision: https://secure.phabricator.com/D14986
Summary:
If you try to pull the hovercard of something you can no longer see (maybe you loaded the page, then the policy changed) there won't be a value in the array here.
(The rest of the code anticipates this possibility.)
Test Plan: Hovered some stuff.
Reviewers: chad
Reviewed By: chad
Differential Revision: https://secure.phabricator.com/D14987
Summary: See Q266.
Test Plan: Created a bulk job, clicked "Details" instead of "Confirm", clicked "Continue" to get back to confirmation dialog.
Reviewers: chad
Reviewed By: chad
Differential Revision: https://secure.phabricator.com/D14985
Summary:
Ref T5155. Swaps Phabricator over to the new first-party S3 client using the v4 authentication API so it works in all regions.
The API requires an explicit region, so the new `amazon-s3.region` is now required. I'll write guidance about this.
Test Plan:
- Uploaded files to S3.
- Migrated ~1GB of files to S3.
- Loaded a bunch of files off S3.
- Browsed around the S3 bucket.
- Deleted a file, verified the data on S3 was destroyed.
- Hit new setup warning.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T5155
Differential Revision: https://secure.phabricator.com/D14982
Summary: Ref T10110. If an image had `width` or `height`, we would accidentally not give it an `$image_class`.
Test Plan:
{F1057988}
{F1057989}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10110
Differential Revision: https://secure.phabricator.com/D14983
Summary: Ref T9828. Mostly just does a minor modernization pass, but also doesn't migrate chunked files since it's not meaningful (they don't have data, directly).
Test Plan: Ran `bin/files migrate` with various flags. Migrated S3 -> Blob and Blob -> S3.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9828
Differential Revision: https://secure.phabricator.com/D14981
Summary:
Ref T8612. If a change affects more than 10K paths + hunks, tell the user it's too big and don't bother trying to write it. We're mostly bounded by INSERTs here.
Also, fix an issue with file upload errors. The keys are real PHP constants, but were accidentally converted to strings in D12797, causing every error to show as "unknown error".
Test Plan: {F1057509}
Reviewers: joshuaspence
Reviewed By: joshuaspence
Maniphest Tasks: T8612
Differential Revision: https://secure.phabricator.com/D14977
Summary: I missed this in the recent icon customziation thing.
Test Plan: Typehaead'ed some projects, saw icons properly.
Reviewers: chad
Reviewed By: chad
Differential Revision: https://secure.phabricator.com/D14975
Summary:
Ref T10010. Ref T5819. General alignment of the stars:
- There were some hacks in Conduit around stripping `fa-...` off icons when reading and writing that I wanted to get rid of.
- We probably have room for a subtitle in the new heavy nav, and using the icon name is a good starting point (and maybe good enough on its own?)
- The project list was real bad looking with redundant tag/names, now it is very slightly less bad looking with non-redundant types?
- Some installs will want to call Milestones something else, and this gets us a big part of the way there.
- This may slightly help to reinforce "tag" vs "policy" vs "group" stuff?
---
I'm letting installs have enough rope to shoot themselves in the foot (e.g., define 100 icons). It isn't the end of the world if they reuse icons, and is clearly their fault.
I think the cases where 100 icons will break down are:
- Icon selector dialog may get very unwieldy.
- Query UI will be pretty iffy/huge with 100 icons.
We could improve these fairly easily if an install comes up with a reasonable use case for having 100 icons.
---
The UI on the icon itself in the list views is a little iffy -- mostly, it's too saturated/bold.
I'd ideally like to try either:
- rendering a "shade" version (i.e. lighter, less-saturated color); or
- rendering a "shade" tag with just the icon in it.
However, there didn't seem to be a way to do the first one right now (`fa-example sh-blue` doesn't work) and the second one had weird margins/padding, so I left it like this for now. I figure we can clean it up once we build the thick nav, since that will probably also want an identical element.
(I don't want to render a full tag with the icon + name since I think that's confusing -- it looks like a project/object tag, but is not.)
Test Plan:
{F1049905}
{F1049906}
Reviewers: chad
Reviewed By: chad
Subscribers: 20after4, Luke081515.2
Maniphest Tasks: T5819, T10010
Differential Revision: https://secure.phabricator.com/D14918
Summary:
Fixes T9323. Two minor fixes:
- On the first commit, don't render a downward line.
- Clean up a 1px spacing issue that had cropped up a while ago when we added icons or something, I think.
Test Plan:
Before:
{F1057248}
After:
{F1057249}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9323
Differential Revision: https://secure.phabricator.com/D14974
Summary:
Fixes T9319. Proxied requests (e.g., in the cluster) for binary files (like images) currently fail because we can not return binary data over Conduit in JSON.
Although Conduit will eventually support binary-safe encodings, a cleaner approach to this is just to return a `filePHID` instead of the raw content. This is generally faster and more flexible, and gives us more opportunities to add caching later.
After making the call, the client pulls the file data separately.
We also no longer need to return a complex data structure because we don't do blame over this call any longer.
Test Plan:
- Viewed images in Diffusion.
- Viewed READMEs in Diffusion.
- Used `bin/differential attach-commit rX Dy` to hit attach pathway.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9319
Differential Revision: https://secure.phabricator.com/D14970
Summary: Ref T9319. Ref T2783. This won't currently work in a future environment where daemons and repositories are not on the same host. Send it over Conduit instead.
Test Plan: Used `bin/differential attach-commit rX Dy` to force attachment, saw valid content pull over Conduit.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T2783, T9319
Differential Revision: https://secure.phabricator.com/D14969
Summary:
Ref T9319. See D14967. As before, this is making a deeply-buried, complex operation easier to test by providing a CLI command.
This adds `bin/differential attach-commit rXnnnn Dnnnn` to pretend that `rXnnnn` was just committed and matched `Dnnnn`.
Test Plan:
- Ran `bin/differential attach-commit X Y` for several different values, saw updates in the UI.
- Faked the message parser to make sure stuff still worked there.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9319
Differential Revision: https://secure.phabricator.com/D14968
Summary: Fixes T9701. I don't want to try to autofix this because destroying the directory could destroy important files, but we can improve the error message.
Test Plan: Faked a failure, ran `repository update X`, got a more tailored error message.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9701
Differential Revision: https://secure.phabricator.com/D14971
Summary:
The old treatment was fairly technical. Give this UI a more human-friendly flow:
- Use language "remove" instead of "disable". We keep the record that the card existed around for auditing/historical purposes, but it is no longer a valid payment method going forward and can not be undone. I think this aligns with user expectation and actual behavior better than "disable".
- Only show active methods on the profile screen.
Test Plan: {F1057153}
Reviewers: chad
Reviewed By: chad
Differential Revision: https://secure.phabricator.com/D14973
Summary:
Ref T9319. When we discover a commit, we sometimes update the corresponding revision with a "this is the actual committed change" diff and send out a link to the changes between review and commit.
This is currently very difficult to test, because it only happens the first time and you have to either go set up a bunch of objects or add a bunch of special casing to the parser to hit the workflow.
I'm making some changes to how it pulls file content. To make those changes easier to test, first start extracting this stuff so the code can be run with `bin/differential extract ...` instead of needing to do a bunch of more complicated setup steps.
Test Plan:
- Ran `bin/differential extract ...` to extract diffs from commits.
- Forced my way through the daemon workflow by faking out a bunch of flags, got a clean extract + attach + update. After this patch, this should rarely be necessary.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9319
Differential Revision: https://secure.phabricator.com/D14967
This is a likely fix for HTTP clones against proxied repositories in the
cluster, although I'm not 100% sure I'm replicating it correctly.
The issue appears to be that we're proxying all the headers, including the
"Transfer-Encoding" header, although the request will already have stripped
any encoding. This might cause us to emit a "chunked" header without a
chunked body.
Auditors: chad
Summary: I was looking at some random un-revisioney repository for most of my testing and missed these.
Test Plan: Viewed blame of a file with some revisions.
Reviewers: chad
Reviewed By: chad
Differential Revision: https://secure.phabricator.com/D14965
Summary: Fixes T2450. If we spend more than 15 seconds in blame, just cut it off.
Test Plan:
- Changed timeout to 0.01 seconds.
- Did blame on a non-highlighted file, got no blame, saw warning.
- Did blame on a highlighted file, got no blame.
- Note: you don't get a warning here because of Ajax stuff. It'd be kind of tricky to add and doesn't seem like a big deal so I'm planning to leave it as-is for now.
Reviewers: chad
Reviewed By: chad
Subscribers: 20after4, chasemp
Maniphest Tasks: T2450
Differential Revision: https://secure.phabricator.com/D14964
Summary:
Ref T2450. This reorganizes code to improve performance.
Mostly, there are a lot of things which are unique per commit (author name, links, short name, etc), but we were rendering them for every line.
This often meant we'd render the same author's name thousands of times. This is slower than rendering it only once.
In 99% of interfaces this doesn't matter, but blame is weird and it's significant on big files.
Test Plan:
Locally, `__phutil_library_map__.php` now has costs of roughly:
- 550ms for main content (from 650ms before the patch).
- 1,500ms for blame content (frrom 1,800ms before the patch).
So this isn't huge, is a decent ~20%-ish performance gain for shuffling some stuff around.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T2450
Differential Revision: https://secure.phabricator.com/D14963
Summary:
Ref T2450. File blame tends to have the same commit a lot of times, and we don't do lookups like this efficiently right now.
In particular, for a file like `__phutil_library_map__.php`, we would issue a query with ~9,000 clauses like this:
```
(repositoryID = 1 AND commitIdentifier LIKE "XYZ%")
```
...but only a few hundred of those identifiers were unique. Instead, issue only one clause per unique identifier.
MySQL also seems to do a little better on "commitIdentifier = X" if we have the full hash, so special case that slightly.
Test Plan:
- Issuing a query for only unique identifiers dropped the cost from 400ms to 100ms locally.
- Swapping to `=` if we have the full hash dropped the cost from 100ms to 75ms locally.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T2450
Differential Revision: https://secure.phabricator.com/D14962
Summary:
Fixes T4366. Two years ago, Facebook put 16,000 files in a directory. Today, the page has nearly loaded.
Paginate large directories.
Test Plan:
- Viewed home and browse views in Git, Mercurial and Subversion.
I put an artificially small page size (5) on home:
{F1055653}
I pushed 16,000 files to a directory and paged through them. Here's the last page, which rendered in about 200ms:
{F1055655}
Our behavior is a bit better than GitHub here, which shows only the first 1,000 files, disables pagination, and can't retrieve history for the files:
{F1055656}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4366
Differential Revision: https://secure.phabricator.com/D14956
Summary: Make sure to subclass the right controller on badges.
Test Plan: arc liberate, make a custom badges edit form.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin
Differential Revision: https://secure.phabricator.com/D14961
Summary:
When looking at a large file in Diffusion:
- disable highlighting if it's huge and show a note about why;
- pick up a few other optimizations.
Test Plan: Locally, this improves the main render of `__phutil_library_map__.php` from 3,200ms to 600ms for me, at the cost of syntax highlighting (we can eventually add view options and let users re-enable it).
Reviewers: chad
Reviewed By: chad
Differential Revision: https://secure.phabricator.com/D14959
Summary:
Ref T2450. Ref T9319. This is still a bit messy, but not quite so bad as it was: instead of using a single call to get both blame information and file content, use `diffusion.blame` for blame information.
This will make optimizations to both blame and file content easier.
Test Plan: Viewed a bunch of blame (color on/off, blame on/off).
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T2450, T9319
Differential Revision: https://secure.phabricator.com/D14958
Summary:
Fixes T2451. Several motivations here, from strongest to weakest:
- Currently, getting blame and file content are closely entwined. This makes fixing T9319 more difficult, and I want to fix it. I want to separate blame from content so there's more flexibility in how we approach this issue.
- This makes pursuing T2450 easier, if it turns out to be a meaningful win.
- If we can get a win on blame performance, we can do `arc blame` eventually if we want.
Test Plan:
- Blamed in SVN, Git and Mercurial.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T2451
Differential Revision: https://secure.phabricator.com/D14957
Summary:
Fixes T10089. This did work at one point, but was broken by D12868, which got too aggressive about mailing members.
We don't want to send mail to all members by default, only those who are subscribed. The parent implementation of `getMailCC()` handles this for us.
Test Plan:
Joined a project as users A and B. Unsubscribed with B. Made an edit.
Before patch: both A and B got mail. After patch: only A got mail.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10089
Differential Revision: https://secure.phabricator.com/D14955
Summary: Adds a basic HeraldAdapter to Phame Blogs and Posts.
Test Plan: Make a Herald rule to CC me on new posts or blogs automatically.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin
Differential Revision: https://secure.phabricator.com/D14897
