NOTE: This branch is not to be deployed as a website, but instead
to be used as a client of a cluster (to have access to the database)
and run the baking script in a unattended mode.
A lot of tweaks all over the place to make the generated HTML suitable
for use in a static HTML with remapped paths to be relative and so on.
NOTE: Because of those tweaks the file rendering and access will not
work as expected when viewed from the website: it will be 404 because
it has different means accessing files than the static HTML. It is not
possible to keep website rendering correct, but it makes it harder to
investigate what parts needs attention without running the baker.
The baking script is found in `scripts/bake/bake_differential.php`
and it receives a single argument which is a directory with the baked
files.
Covers the following aspects:
- File attachments
- Image previews
- Pastes
The user avatars are replaced with the default user avatar.
Was a mistake in the way how key de-duplication and non-phabricator-managed
commit access was dealt with.
For now simple fix: make sure user list is unique. Proper fix would require
something more sophisticated.
This causes an extra file read, but the benefit is that it is easier
to compare data on the file system, quickly see what did actually change
and what did not.
Implements following rules:
- Users
- Users of any project
- Users of all projects
- Administrators
- Signers
The 'If No Rules Match' the access is implicitly considered to
be 'DENY'.
It is not possible to control access based on the Moon phase.
Implements following rules:
- Users
- Users of any project
- Users of all projects
- Administrators
- Signers
The 'If No Rules Match' the access is implicitly considered to
be 'DENY'.
It is not possible to control access based on the Moon phase.
Summary: Ref T13395. Since there's very little code which really makes sense in "scripts/", I've moved most of it to other places.
Test Plan: Ran `bin/phd`.
Maniphest Tasks: T13395
Differential Revision: https://secure.phabricator.com/D20994
Summary: Ref T13395. Moves a small amount of remaining "libphutil/" code into "phabricator/" and stops us from loading "libphutil/".
Test Plan: Browsed around; there are likely remaining issues.
Maniphest Tasks: T13395
Differential Revision: https://secure.phabricator.com/D20981
Summary:
Ref T13436. There's no real security value to doing this comparison, it just wards off evil "security researchers" who get upset if you ever compare two strings with a non-constant-time algorithm.
In practice, SSH public keys are pretty long, pretty public, and have pretty similar lengths. This leads to a relatively large amount of work to do constant-time comparisons on them (we frequently can't abort early after identifying differing string length).
Test Plan: Ran `bin/ssh-auth --sshd-key ...` on `secure` with ~1K keys, saw runtime drop by ~50% (~400ms to ~200ms) with `===`.
Maniphest Tasks: T13436
Differential Revision: https://secure.phabricator.com/D20875
Summary: Depends on D20873. Ref T13436. Allow callers to configure "bin/ssh-auth --sshd-key %k" as an "AuthorizedKeysCommand"; if they do, and we recognize the key, emit just that key in the output.
Test Plan:
- Used `git pull` locally, still worked fine.
- Instrumented things, saw the public key lookup actually work and emit a single key.
- Ran without "--sshd-key", got a full key list as before.
Maniphest Tasks: T13436
Differential Revision: https://secure.phabricator.com/D20874
Summary:
Ref T13436. Historically, this script could be used with a forked copy of "sshd" to do lower-cost per-key auth.
Relatively modern "sshd" supports "%f" to "AuthorizedKeysCommand", which effectively moots this.
Users have never been instructed to use this script for anything, and we moved away from this specific patch to "sshd" some time ago.
Test Plan: Grepped for "ssh-auth-key", no hits.
Maniphest Tasks: T13436
Differential Revision: https://secure.phabricator.com/D20873
Summary:
Fixes T13382. Depends on D20724. These ancient scripts are no longer necessary since we've had a smooth web-based onboarding process for a long time.
I retained `bin/user empower` and `bin/user enable` for recovering from situations where you accidentally delete or disable all administrators. This is normally difficult, but some users are industrious.
Test Plan: Grepped for `accountadmin` and `add_user.php`, found no more hits.
Maniphest Tasks: T13382
Differential Revision: https://secure.phabricator.com/D20725
Summary:
Ref T13382.
- Remove "bin/people profileimage" which previously generated profile image caches but now feels obsolete.
- Replace it with "bin/user", with "enable" and "empower" flows. This command is now focused on regaining access to an install after you lock your keys inside.
- Document the various ways to unlock objects and accounts from the CLI.
Test Plan:
- Ran `bin/user enable` and `bin/user empower` with various flags.
- Grepped for `people profileimage` and found no references.
- Grepped for `bin/people` and found no references.
- Read documentation.
Maniphest Tasks: T13382
Differential Revision: https://secure.phabricator.com/D20724
Summary: Depends on D20069. Ref T13232. This is a very, very weak dependency and we can reasonably polyfill it.
Test Plan: Grepped for `iconv` in libphutil, arcanist, and Phabricator.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13232
Differential Revision: https://secure.phabricator.com/D20070
Summary: If you go through the `accountadmin` flow and change nothing, you get an exception about the transaction not having any effect. Instead, let the `applyTransactions` call continue even on no effect.
Test Plan: Ran `accountadmin` without changing anything for an existing user. No longer got an exception about no-effect transactions.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin
Differential Revision: https://secure.phabricator.com/D20009
Summary:
Ref PHI1027. Currently, `PhabricatorUser` has a couple of mail-related methods which shouldn't really be there in the long term. Immediately, I want to make some adjusments to the welcome email.
Move "Welcome" mail generation to a separate class and consolidate all the error handling. (Eventually, "invite" and "verify address" email should move to similar subclasses, too.) Previously, a bunch of errors/conditions got checked in multiple places.
The only functional change is that we no longer allow you to send welcome mail to disabled users.
Test Plan:
- Used "Send Welcome Mail" from profile pages to send mail.
- Hit "not admin", "disabled user", "bot/mailing list" errors.
- Used `scripts/user/add_user.php` to send welcome mail.
- Used "Create New User" to send welcome mail.
- Verified mail with `bin/mail show-outbound`. (Cleaned up a couple of minor display issues here.)
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D19989
Summary:
After T13217 import_repository_symbols.php was showing a lot of warnings, using %LQ fixes that.
I'm aware, that there are changes planned to the whole managing the symbols complex but until then less warnings are nice.
Test Plan: No more warnings when updating symbols
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D19962
Summary:
Ref T920. Over time, mail has become much more complex and I think considering "mail", "sms", "postcards", "whatsapp", etc., to be mostly-the-same is now a more promising avenue than building separate stacks for each one.
Throw away all the standalone SMS code, including the Twilio config options. I have a separate diff that adds Twilio as a mail adapter and functions correctly, but it needs some more work to bring upstream.
This permanently destroys the `sms` table, which no real reachable code ever wrote to. I'll call this out in the changelog.
Test Plan:
- Grepped for `SMS` and `Twilio`.
- Ran storage upgrade.
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T920
Differential Revision: https://secure.phabricator.com/D19939
Summary: Ref T13216. See D19666. It's currently tricky to profile Herald test runs since you have to submit a form and repeating them is a bit of a mess. Provide a simple CLI wrapper so we can use `--xprofile`. This is also maybe nice-to-have if we're ever debugging anything here.
Test Plan: Ran `bin/herald test --object ... --type ...` and got a sensible looking transcript in the UI.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13216
Differential Revision: https://secure.phabricator.com/D19806
Summary: Depends on D19796. Simplify some timing code by using phutil_microseconds_since() instead of duplicate casting and arithmetic.
Test Plan: Grepped for `1000000` to find these. Pulled, pushed, made a conduit call. This isn't exhaustive but it should be hard for these to break in a bad way since they're all just diagnostic.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D19797
Summary:
Depends on D19779. Ref T13216. The push logs currently record the "hostWait", which is roughly "locking + subprocess cost". We also record locking separately, so we can figure out "subprocess cost" alone by subtracting the lock costs.
However, the subprocess (normally `git receive-pack`) runs hooks, and we don't have an easy way to figure out how much time was spent doing actual `git` stuff vs spent doing commit hook processing. This would have been useful in diagnosing at least one recent issue.
Track at least a rough hook cost and record it in the push logs.
Test Plan: Pushed to a repository, saw a reasonable hook cost appear in the database table.
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T13216
Differential Revision: https://secure.phabricator.com/D19780
