Summary:
Differential uses a byte-oriented linewrap algorithm. Instead, use a
character-oriented one which will handle utf-8 properly.
This implies a very slightly performance hit but we only run this code for lines
which need to wrap, and the results get cached. It took about ~2.5ms for the
test file on my machine. I'll keep an eye on it but I think it's currently a
manageable cost.
Test Plan:
Diffed this file: https://secure.phabricator.com/P43
...and got it to render like this:
https://secure.phabricator.com/file/info/PHID-FILE-331ac241bede705b193b/
To do so, I had to disable the un-utf8 block which we can't actually do yet
because of intraline diff, but it shows that once we can get rid of that it
works completely correctly. It will "sort of" work in the meantime (nothing
terrible happens).
Reviewers: jungejason, aran, tuomaspelkonen
CC: aran, epriestley
Differential Revision: 513
Summary:
When you type "5 PM", it should mean "5 PM in your local time", not "5 PM in the
server's time". Use locale-aware functions to read and write times.
Also a couple of usability tweaks.
Test Plan:
Swapped my time between America/Los_Angeles and America/New_York and created a
bunch of countdowns.
Reviewed By: jungejason
Reviewers: hsb, toulouse, jungejason, tuomaspelkonen, aran
CC: aran, jungejason
Differential Revision: 531
Summary:
Replace some more date() calls with locale-aware calls.
Also, at least on my system, the DateTimeZone / DateTime stuff didn't actually
work and always rendered in UTC. Fixed that.
Test Plan:
Viewed daemon console, differential revisions, files, and maniphest timestamps
in multiple timezones.
Reviewed By: toulouse
Reviewers: toulouse, fratrik, jungejason, aran, tuomaspelkonen
CC: aran, toulouse
Differential Revision: 530
Summary:
create the indexer for commit
Test Plan:
run the reindex_one_commit.php against one existing commit
and it is returned in search result on the webpage; run parse_one_commit
against another commit; modified reindex_everything.php to let it only
parse one commit after loading all commits.
Reviewed By: epriestley
Reviewers: epriestley, aran
CC: aran, jungejason, epriestley, debow
Differential Revision: 490
Summary:
Provides basic Remarkup support for @mentions. No application integration yet so
these aren't terribly useful until that happens.
Test Plan:
https://secure.phabricator.com/file/view/PHID-FILE-83d68e7af6085ae928df/
Reviewers: tomo, mroch, jsp
Commenters: tomo
CC: aran, tomo, epriestley
Differential Revision: 517
Summary: This depends on D513 and D514. Those diffs make the display algorithms safe with respect to mutating utf8, so we no longer need to repair potentially invalidate utf8 sequences with this hack.
Test Plan: grepped for calls to this method
Reviewers: jungejason, aran, tuomaspelkonen
CC:
Differential Revision: 515
Summary: Differential uses a byte-oriented linewrap algorithm. Instead, use a character-oriented one which will handle utf-8 properly.
This implies a very slightly performance hit but we only run this code for lines which need to wrap, and the results get cached. It took about ~2.5ms for the test file on my machine. I'll keep an eye on it but I think it's currently a manageable cost.
Test Plan: Diffed this file: https://secure.phabricator.com/P43
...and got it to render like this: https://secure.phabricator.com/file/info/PHID-FILE-331ac241bede705b193b/
To do so, I had to disable the un-utf8 block which we can't actually do yet because of intraline diff, but it shows that once we can get rid of that it works completely correctly. It will "sort of" work in the meantime (nothing terrible happens).
Reviewers: jungejason, aran, tuomaspelkonen
CC:
Differential Revision: 513
Summary:
Add users to the search results. I need to follow this up with a patch to make
the search results stop being terrible. I'll do that.
Test Plan:
Searched for users, ran "reindex_all_users.php"
Reviewed By: jungejason
Reviewers: tomo, jungejason, aran
CC: aran, jungejason
Differential Revision: 508
Summary:
Diffs with missing context don't render properly in the "ignore all whitespace"
algorith, so don't try to use it. These diffs can occur if someone creates a
diff via the web interface, for example, or if they muck around in their copy of
'arc'.
See D473, T246 (a problem with D473), rPe5bb756b5191720 (revert of D473) and
T231.
Test Plan:
Viewed a diff with missing context from the web interface. Verified normal diffs
still rendered with all whitespace ignored.
Reviewed By: fratrik
Reviewers: jungejason, aran, tuomaspelkonen, fratrik
Commenters: jungejason
CC: aran, epriestley, fratrik, jungejason
Differential Revision: 500
Summary:
See comments. I think this will fix the issue, where we end up handling off
garbage to htmlspecialchars() after highlighting a file we've stuck full of \0
bytes.
The right fix for this is to make wordwrap and intraline-diff utf8 aware and
throw this whole thing away. I'll work on that but I think this fixes the
immediate issue.
Test Plan:
diffed the file with a UTF-8 quote in it and got a reasonable render in
Differential
Reviewed By: jungejason
Reviewers: jungejason, aran, tuomaspelkonen
CC: aran, jungejason
Differential Revision: 504
Summary:
I didn't grep carefully enough when renaming this to 'loadProject' in D485. See
task. I am terrible at the programmings.
Test Plan:
Hit the "Edit project" view, grepped more thoroughly. :P
Reviewed By: tomo
Reviewers: tomo, toulouse, cadamo
CC: aran, tomo
Differential Revision: 503
Summary:
See T251, where gregprice correctly argues that we need both:
None of the other people on the thread will have seen that message, so it
seems
like a lot of clients would put the server's message in a new thread. In
general, I think you want the References: header to mention every ancestor
message in the thread that you know about, because that's how MUAs keep a
thread
together in the face of missing some of its messages.
Test Plan:
Sent a reply email locally, got a response with both Message-IDs in
"references".
Reviewed By: rm
Reviewers: gregprice, rm
Commenters: gregprice
CC: aran, gregprice, epriestley, rm
Differential Revision: 499
Summary:
See T251. In Gmail, conversations split if you reply to them and the next email
does not "In-Reply-To" your message ID. When an action is triggered by an email,
carry its Message-ID through the stack and use it for "In-Reply-To" and
"References" on the subsequent message.
Test Plan:
Live-patched phabricator.com and replied to a Maniphest thread in Gmail without
disrupting the thread. Locally replied to Maniphest and Differential threads and
verified Message-ID was carried across the reply boundary.
Reviewed By: rm
Reviewers: tcook, jungejason, aran, tuomaspelkonen, rm
CC: aran, epriestley, rm
Differential Revision: 498
Summary:
D477 added functionality to the project list view but had a couple of
performance issues that I missed in review, because it took the query count for
the page from around 3 to as many as 300, including up to 100 heavyweight search
index queries.
This fixes the two simple N+1 query problems. This general pattern of data
access often occurs:
COUNTEREXAMPLE
$cats = load_cats();
foreach ($cats as $cat) {
$cats_hats = load_hats_for_cat($cat);
// ...
}
But this issues "N+1" queries, i.e. if you load 100 cats you issue 101 queries.
It is faster to group the queries instead:
$cats = load_cats();
$hats = load_all_hats_for_these_cats($cats);
foreach ($cats as $cat) {
$cats_hats = $hats[$cat->getID()];
}
MySQL can execute one query which returns all the results much faster than 100
queries which return one result, especially if the database is not local (i.e.,
over the network).
However, this doesn't save a ton of time. The bigger issue is that I didn't have
the right keys on the relationship tables in the search engine. This adds them,
and reduces the search engine lookup cost from 25-80ms (for
secure.phabricator.com) down to 1-3ms.
I still probably want to get this out of the loop at some point but it's okay
for now and the page loads in a few ms rather than taking more than a second.
Test Plan:
Used "services" tab, "xhprof" and "EXPLAIN" to analyze page performance. I
measured these changes:
- Query count: 1 + (3 * N projects) -> 3 + (N projects) (e.g., 301 -> 103)
- Total time spent querying, ignoring search indexes: 40ms (local.aprhont.com)
-> 20ms (local.aphront.com)
- Cost for search index query: 25-80ms (secure.phabricator.com) -> 1-3ms
Reviewed By: cadamo
Reviewers: cadamo, aran, jungejason, tuomaspelkonen
CC: aran, cadamo, epriestley
Differential Revision: 485
Summary:
If you "chmod +x" a file and then generate a diff and click "show 20 lines" on
that diff, you get another janky copy of the property table. Render the property
table only for top-level rendering requests.
This started happening after D409, which fixed the far-more-obvious bug of these
things never showing up. We must have changed the logic at some point since this
side effect was surprising to me. :P
Test Plan:
Created a diff with changes and +x, clicked "show 20 lines".
- Original diff had file property header table thing, showing the +x.
- New context brought in by "show 20 lines" didn't have it anymore.
Reviewed By: jungejason
Reviewers: aran, jungejason, tuomaspelkonen
CC: aran, epriestley, jungejason
Differential Revision: 482
Summary:
This ended up being pretty hard to see, make it a bit easier.
Test Plan:
Focused things using the keyboard reticle.
Reviewed By: tomo
Reviewers: tomo, moskov, jungejason, tuomaspelkonen, aran
CC: aran, epriestley, tomo
Differential Revision: 483
Summary:
For some reason, Webkit parses the completely made-up "background-position-y"
property. Firefox does not. Use a real property instead of a creative one that
doesn't exist.
Test Plan:
Hovered over "Phabricator" logo in Firefox, Safari.
Reviewed By: codeblock
Reviewers: codeblock, aran, jungejason, tuomaspelkonen
CC: aran, codeblock
Differential Revision: 484
Summary:
Without this, user creation throws an exception when trying to insert NULL into
a non-NULL field.
Test Plan:
Created a new user.
Reviewed By: fratrik
Reviewers: fratrik, toulouse, aran, jungejason, tuomaspelkonen
CC: aran, fratrik
Differential Revision: 480
Summary: Added some change on the project's list view, to show information about
active tasks, population, etc. Also modified the "profile view", and added a class "PhabricatorProfileView" to render the profile, both on projects and users.
Test Plan: play around the project directory :)
Reviewers: epriestley ericfrenkiel
CC:
Differential Revision: 477
Summary:
create the event
Test Plan:
checked the timeline event was consumed successfully by
facebook daemon when the revision is created from the webUI or arc
command line.
Reviewed By: epriestley
Reviewers: epriestley, aran
CC: aran, epriestley
Differential Revision: 478
Summary:
Provides a new workflow for making it non-horrible to install certificates.
Basically you run "arc install-certificate" and then copy/paste a short token
off a webpage and it does the ~/.arcrc edits for you.
Test Plan:
Installed certificates, used bad tokens, hit rate limiting.
Reviewed By: aran
Reviewers: aran, jungejason, tuomaspelkonen
CC: aran
Differential Revision: 460
Summary:
The owners of a revision are only really the reviewers when the revision is in
NEEDS_REVIEW.
Also build a raw indexed document viewer so you can look at the index of a
document from the web interface.
Finally, reindex revisions when comments are added, not just when the revision
itself is edited.
Test Plan:
Toggled abandon/reclaim on a revision and verified the relationships indexed
properly.
Reviewed By: jungejason
Reviewers: aran, jungejason, tuomaspelkonen
CC: aran, epriestley, jungejason
Differential Revision: 470
Summary:
The MySQL MyISAM fulltext backend defaults to a minimum length of 4, but many
3-letter words are useful search terms. Provide instructions for configuring
MySQL to reduce the auto-stopword length.
Test Plan:
Followed instructions, searched for a 3-letter term and got a hit.
Reviewed By: jungejason
Reviewers: aran, tuomaspelkonen, jungejason
CC: aran, jungejason
Differential Revision: 471
Summary:
'Create Diff' with whitespace mode 'ignore-all' is borken, because the
line numbers get mixed up when creating a second diff for the
whitespace changes.
This should be fixed correctly at some point, but currently the
whitespace 'ignore-all' says 'Huge mess' in the comments and I didn't
want to make the mess any bigger.
Test Plan:
Tested that 'Create Diff' showed the diff correctly and a previously
created diff looked correct once the cache was disabled.
Reviewed By: epriestley
Reviewers: epriestley, jungejason
CC: aran, epriestley
Differential Revision: 473
Summary:
Allows user-configurable timezones. Adds a preference panel, and migrates to the
new date rendering in easily-modified areas of the code. ***In progress***.
Test Plan:
Check database to make sure the field is being changed when the settings are
changed; check affected views to see how they render times.
Reviewed By: epriestley
Reviewers: epriestley
CC: aran, epriestley, toulouse
Differential Revision: 475
Summary:
Paste was cutting off lines when they extended past the width of the screen.
Also change a pair of double quotes to single quotes, to follow convention.
Test Plan:
Tried this in Firefox 4 and Chrome.
Reviewers:
epriestley, toulouse
CC:
Differential Revision: 469
Summary:
Make it slightly more clear how to go about this.
Test Plan:
Generated and reviewed the documentation.
Reviewed By: moskov
Reviewers: moskov, aran, jungejason, tuomaspelkonen
CC: aran, moskov
Differential Revision: 467
Differential
Summary:
Make some display stuff more consistent.
Test Plan:
Looked at a task and a revision.
Reviewed By: jungejason
Reviewers: aran, jungejason, tuomaspelkonen
CC: aran, jungejason
Differential Revision: 462
Summary:
Permit "j" and "k" to cycle through individual changeblocks, similar to how this
feature works in ReviewBoard. This still needs a bunch of refinement but it's
getting closer to being useful.
Also moved reticle underneath the table so you can click links through it (derp
derp).
Test Plan:
Used "j" and "k" to cycle through individual changes.
Reviewed By: aran
Reviewers: aran, jungejason, tuomaspelkonen
CC: moskov, aran, epriestley
Differential Revision: 426
Summary:
Allow duplicate tasks to be selected and merged in Maniphest.
I didn't create a separate transaction type for this because that implies a
bunch of really complicated rules which I don't want to sort out right now
(e.g., do we need to do cycle detection for merges? If so, what do we do when we
detect a cycle?) since I think it's unnecessary to get right for the initial
implementation (my Tasks merge implementation was similar to this and worked
quite well) and if/when we eventually need the metadata to be available in a
computer-readable form that need should inform the implementation.
Plenty of room for improvement here, of course.
Test Plan:
Merged duplicate tasks, tried to perform invalid merge operations (e.g., merge a
task into itself).
Tested existing attach workflows (task -> revision, revision -> task).
Reviewed By: aran
Reviewers: tuomaspelkonen, jungejason, aran
CC: anjali, aran
Differential Revision: 459
Summary:
See attached tasks. See D459 for the ability to merge tasks.
Test Plan:
Looked at posted and unposted inline comments.
Reviewed By: aran
Reviewers: edward, viyer, aran, jungejason, tuomaspelkonen
CC: aran, epriestley
Differential Revision: 461
Summary:
For some unknown reasons, our test framework fails to update test
results occasionally. Instead of throwing a nasty exception, log
the error message. This will help us to figure out the real reason
behind the error.
Test Plan:
arc lint.
Reviewed By: epriestley
Reviewers: jungejason, epriestley
CC: epriestley, aran
Differential Revision: 463
