Summary:
Make it possible not to allow anyone to edit Diffusion identities.
Make it possible not to allow anyone to view other users' email addresses.
Closes T15443
Test Plan:
* As an admin, go to `/applications/view/PhabricatorDiffusionApplication/` and see new policy "Can Edit and View Identities" set to "All Users" (as implicitly before)
* As an admin, go to `/applications/view/PhabricatorDiffusionApplication/` and change "Can Edit and View Identities" from "All Users" to "Administrators"
* As a non-admin, go to `/diffusion/identity/` and try to select the disabled "Create Identity" button; get an error message clicking it due to lack of permissions
* Given there is at least one identity defined, as a non-admin, go directly to `/diffusion/identity/view/1/` and get "You do not have permission to view this object."
* Given there is at least one identity defined, as a non-admin, go directly to `/diffusion/identity/edit/1/` and get "You do not have permission to view this object."
* As a non-admin, go directly to `/diffusion/identity/edit/form/default/` and get "You do not have permission to edit this object."
* As a non-admin, go directly to `/diffusion/identity/` and get "No Identities found." instead of seeing the existing identities listed.
* As an admin, go to `/diffusion/identity/` and still see the existing identities listed.
* As an admin, go to `/diffusion/identity/`, select "Create Identity" to go to `/diffusion/identity/edit/` and see the "Create Identity" page (though broken; see T15453)
* As an admin, go to `/diffusion/identity/view/1/` and still see the existing identity.
* As an admin, go to `/diffusion/identity/edit/1/` and successfully edit the existing identity.
Reviewers: O1 Blessed Committers, speck, valerio.bozzolan
Reviewed By: O1 Blessed Committers, speck, valerio.bozzolan
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15443
Differential Revision: https://we.phorge.it/D25450
Summary:
When throwing an exception related to output provided by `git cat-file`, include the repository monogram to allow potentially debugging in Git.
Closes T15661
Test Plan:
Unclear. Basically: "have a broken Git repository in Diffusion".
(However this patch changes a message only shown in case of an exception, so in the worst case we'd break the exception via an exception.)
Reviewers: O1 Blessed Committers, speck
Reviewed By: O1 Blessed Committers, speck
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15661
Differential Revision: https://we.phorge.it/D25460
Summary:
Left-over from rPa6b550ba0394284441ee55d11e276a05eb568ad9
Closes T15655
Test Plan: Carefully read the source code file to look out for another appearance; after hours of fruitless searching get slightly disappointed.
Reviewers: O1 Blessed Committers, avivey
Reviewed By: O1 Blessed Committers, avivey
Subscribers: tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15655
Differential Revision: https://we.phorge.it/D25451
Summary:
Fix the following error you may encounter in production:
Too few arguments to function DiffusionDiffQueryConduitAPIMethod::getDefaultParser(), 0 passed in /var/www/phorge/src/applications/diffusion/conduit/DiffusionDiffQueryConduitAPIMethod.php on line 156 and exactly 1 expected
It was caused by the lack of a new mandatory parameter in a method.
Closes T15649
Test Plan: Check with your big eyes that a ConduitAPIRequest is passed to getDefaultParser().
Reviewers: avivey, O1 Blessed Committers, speck
Reviewed By: O1 Blessed Committers, speck
Subscribers: speck, tobiaswiese, Matthew, Cigaryno
Maniphest Tasks: T15649
Differential Revision: https://we.phorge.it/D25444
Summary:
This broadens the typeahead datasource search for repos. Before this patch a repository named "Alligator Simulator" would not be found with the search string "simu...". This is patched with the ferret engine search and indexing features.
See T15583
Test Plan: Create repositories with titles with 2 or more words. Search for these repos with the global typeahead search. The search term should begin with the second/third/n-th word of the repo title.
Reviewers: O1 Blessed Committers, avivey
Reviewed By: O1 Blessed Committers, avivey
Subscribers: avivey, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15583
Differential Revision: https://we.phorge.it/D25430
Summary:
If the user specifies a text encoding via the "View Options" dropdown, respect this choice.
Ref Q68.
Test Plan: Play with the Encoding button in the view
Reviewers: O1 Blessed Committers, valerio.bozzolan
Reviewed By: O1 Blessed Committers, valerio.bozzolan
Subscribers: tinloaf, speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Differential Revision: https://we.phorge.it/D25360
Summary:
Make sure that `$epochs[]` is never empty but contains `0` so `min($epochs)` will not complain.
Closes T15623
Test Plan: Carefully read the code (as I am unaware of reproduction steps).
Reviewers: O1 Blessed Committers, avivey
Reviewed By: O1 Blessed Committers, avivey
Subscribers: tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15623
Differential Revision: https://we.phorge.it/D25417
Summary:
This call is preventing users to browse Subversion repositories.
Using strlen() to check string validity is deprecated since PHP 8.1, phorge adopts phutil_nonempty_string() as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
Fix T15608
Test Plan:
- Sign in
- Open a Diffusion SVN repository
- You should see the repository instead of getting a Runtime Exception
Reviewers: O1 Blessed Committers, Sten, avivey
Reviewed By: O1 Blessed Committers, Sten, avivey
Subscribers: Sten, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15608
Differential Revision: https://we.phorge.it/D25398
Summary:
With PHP 8.1+ it was not possible to view a commit from subversion repositories.
Indeed, if the commit user and/or email is not properly defined, strlen(null) is called, causing a deprecation warning, elevated to exception.
Using strlen() to check string validity is deprecated since PHP 8.1. Phorge adopts phutil_nonempty_string() as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
Fix T15609
Test Plan:
- Apply D25397 and D25398
- Sign in
- Open a diffusion SVN repository
- Open a commit without user name and or email
- You should not see the same Runtime Exception (unfortunately, there is another one as described in T15610)
Reviewers: O1 Blessed Committers, valerio.bozzolan
Reviewed By: O1 Blessed Committers, valerio.bozzolan
Subscribers: tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15609
Differential Revision: https://we.phorge.it/D25399
Summary:
These calls are preventing users to browse subversion/mercurial repositories in PHP 8.1+.
Indeed, a similar bug affecting git repositories was already addressed in another commit
(rP6b8ec50148909938850b5acfd11725ae23a8e31b). This commit harmonize both DiffusionSvnRequest
and DiffusionMercurialRequest with DiffusionGitRequest
Fix T15607
Test Plan:
- Sign in
- Open a diffusion SVN/Mercurial repository
- You should not get a RuntimeException concerning this preg_match call
Reviewers: O1 Blessed Committers, Sten, valerio.bozzolan
Reviewed By: O1 Blessed Committers, Sten, valerio.bozzolan
Subscribers: Sten, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15607
Differential Revision: https://we.phorge.it/D25397
Summary:
Doing a git push into a fresh repository over https fails with a PHP 8.1 strlen(null) error.
This fixes it.
Fixes T15569
Test Plan:
* Create an empty repository on a PHP 8.1 Phorge instance
* Clone it
* Add some data
* git push
Reviewers: O1 Blessed Committers, speck
Reviewed By: O1 Blessed Committers, speck
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15569
Differential Revision: https://we.phorge.it/D25361
Summary:
When viewing the top level of a diffusion repository, $request->getPath() returns null, which then causes trim() to throw an error.
Fix is to update DiffusionRequest getPath() such that it returns '' instead of null, thus fixing it's return type.
Fixes T15522
Test Plan: Go to https://my.phorge.site/source/myrepo
Reviewers: O1 Blessed Committers, valerio.bozzolan, avivey
Reviewed By: O1 Blessed Committers, valerio.bozzolan, avivey
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15522
Differential Revision: https://we.phorge.it/D25329
Summary:
Fix PHP 8.1 strlen(null) in DiffusionView linkBrowse().
Fixes T15542
Test Plan:
# Go into a diff (eg https://my.phorge.site/D1234)
# Click on the Revision Contents - History tab
# Click on a base link (the short git revision ID)
Reviewers: O1 Blessed Committers, avivey
Reviewed By: O1 Blessed Committers, avivey
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15542
Differential Revision: https://we.phorge.it/D25340
Summary:
This change implements some messages that were only present
from this class:
DifferentialRevisionHasCommitEdgeType.php
But not in its inverse class:
DiffusionCommitHasRevisionEdgeType.php
In short:
| Before | After |
|-----------|-----------|
| {F313282} | {F313281} |
Closes T15314
Test Plan:
Visit a Diffusion commit and use the sidebar menu "Edit Related Objects" to:
- add 1 revision
- add 2 revisions
- remove 1 revision
- remove 2 revisions
- add 1 revision, remove 1 revision
- add 2 revisions, remove 2 revisions
Check that you don't see phrases like "added an edge" anymore.
Reviewers: O1 Blessed Committers, avivey
Reviewed By: O1 Blessed Committers, avivey
Subscribers: speck, tobiaswiese, Matthew, Cigaryno
Maniphest Tasks: T15314
Differential Revision: https://we.phorge.it/D25306
Summary:
When a 'git pull' is done to an https git URL, the $_SERVER variables PHP_AUTH_USER and PHP_AUTH_PW will be unset, causing PHP 8.1 to throw strlen(null) errors.
This update fixes the issue by defaulting the values to '', which results in $have_user and $have_pass having false values as desired.
Fixes T15520
Test Plan: arc unit
Reviewers: O1 Blessed Committers, valerio.bozzolan
Reviewed By: O1 Blessed Committers, valerio.bozzolan
Subscribers: avivey, speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15520
Differential Revision: https://we.phorge.it/D25327
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=diffusionEmptyRepoPatternSearch, ref.master=97e163187418, ref.diffusionEmptyRepoPatternSearch=97e163187418), phorge(head=master, ref.master=108cbcd09bd3)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/controller/DiffusionTagListController.php:28]
```
Closes T15461
Test Plan: After applying this line, get the expected `CommandException: Command failed with error #128!` due to ` fatal: detected dubious ownership in repository at '/var/repo/1'` (plus a `RuntimeException: file_exists(): Passing null to parameter #1 ($filename) of type string is deprecated` as a bonus), instead of the previous RuntimeException about `strlen()`.
Reviewers: O1 Blessed Committers, speck
Reviewed By: O1 Blessed Committers, speck
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15461
Differential Revision: https://we.phorge.it/D25285
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=master, ref.master=97e163187418), phorge(head=diffusionCallsignHistory, ref.master=9b99123ff931, ref.diffusionCallsignHistory=d5fb6702a49a)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/controller/DiffusionCommitController.php:31]
```
Closes T15466
Test Plan: After applying this change, going to an existing commit URL like `/diffusion/XYZ/commit/999a2c7d1ff8` shows the exception covered by T15464 instead, as expected.
Reviewers: O1 Blessed Committers, speck
Reviewed By: O1 Blessed Committers, speck
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15466
Differential Revision: https://we.phorge.it/D25290
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=master, ref.master=97e163187418), phorge(head=diffusionBrowseCode, ref.master=108cbcd09bd3, ref.diffusionBrowseCode=108cbcd09bd3)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/controller/DiffusionBrowseController.php:25]
```
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=master, ref.master=97e163187418), phorge(head=diffusionBrowseCode, ref.master=108cbcd09bd3, ref.diffusionBrowseCode=108cbcd09bd3)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/conduit/DiffusionBrowseQueryConduitAPIMethod.php:40]
```
Closes T15462
Test Plan: After applying these two changes, get next expected exceptions about `RuntimeException: file_exists(): Passing null to parameter #1 ($filename) of type string is deprecated` and `CommandException: Command failed with error #128! fatal: detected dubious ownership in repository at '/var/repo/1'`, but no RuntimeException about `strlen()` anymore.
Reviewers: O1 Blessed Committers, speck
Reviewed By: O1 Blessed Committers, speck
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15462
Differential Revision: https://we.phorge.it/D25286
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=master, ref.master=97e163187418), phorge(head=diffusionBrowseCommit, ref.master=9b99123ff931, ref.diffusionBrowseCommit=9b99123ff931)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/controller/DiffusionCommitController.php:925]
```
Closes T15464
Test Plan: After applying this change, selecting the last commit on http://phorge.localhost/diffusion/query/all/ correctly renders the commit page (e.g. `/R1:1234567890abcdef1234567890abcdef12345678`)
Reviewers: O1 Blessed Committers, speck
Reviewed By: O1 Blessed Committers, speck
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15464
Differential Revision: https://we.phorge.it/D25288
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=master, ref.master=97e163187418), phorge(head=diffusionBrowseCode, ref.master=108cbcd09bd3, ref.diffusionBrowseCode=015ffef14b0c)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/conduit/DiffusionHistoryQueryConduitAPIMethod.php:50]
```
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=master, ref.master=97e163187418), phorge(head=diffusionBrowseCode, ref.master=108cbcd09bd3, ref.diffusionBrowseCode=015ffef14b0c)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/conduit/DiffusionHistoryQueryConduitAPIMethod.php:50]
```
Closes T15463
Test Plan: After applying these two changes, get next expected exceptions about `RuntimeException: file_exists(): Passing null to parameter #1 ($filename) of type string is deprecated` and `CommandException: Command failed with error #128! fatal: detected dubious ownership in repository at '/var/repo/1'`, but no RuntimeException about `strlen()` anymore.
Reviewers: O1 Blessed Committers, speck
Reviewed By: O1 Blessed Committers, speck
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15463
Differential Revision: https://we.phorge.it/D25287
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=master, ref.master=b325304b6e52), phorge(head=HgViewBranches, ref.master=cb938d869c3f, ref.HgViewBranches=cb938d869c3f)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/conduit/DiffusionBranchQueryConduitAPIMethod.php:83]
```
Closes T15447
Test Plan: After applying this change, going to `/diffusion/1/branches/default/` shows an expected FilesystemException instead of the previous RuntimeException.
Reviewers: O1 Blessed Committers, speck
Reviewed By: O1 Blessed Committers, speck
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15447
Differential Revision: https://we.phorge.it/D25279
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=master, ref.master=b325304b6e52), phorge(head=SvnRepoMain, ref.master=cb938d869c3f, ref.SvnRepoMain=cb938d869c3f)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/controller/DiffusionController.php:156]
```
Closes T15446
Test Plan: After applying this change, going to a newly created SVN repository at `/diffusion/1/`, the page renders, showing the repository as Inactive and also "Empty Repository - This repository does not have any commits yet." as expected.
Reviewers: O1 Blessed Committers, speck
Reviewed By: O1 Blessed Committers, speck
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15446
Differential Revision: https://we.phorge.it/D25278
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=master, ref.master=e4fd31ec024e), phorge(head=D25241, ref.master=b1edfea09bad, ref.D25241=b1edfea09bad)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/controller/DiffusionBranchTableController.php:29]
```
```
EXCEPTION: (RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=master, ref.master=e4fd31ec024e), phorge(head=D25241, ref.master=b1edfea09bad, ref.D25241=b1edfea09bad)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<phorge>/src/applications/diffusion/conduit/DiffusionBranchQueryConduitAPIMethod.php:33]
```
Closes T15414
Test Plan: Applied these two changes and at least got to `(FilesystemException) Filesystem path "/var/repo/1/" does not exist.` as an error message, instead of being stuck at `(RuntimeException) strlen(): Passing null to parameter #1 ($string) of type string is deprecated`.
Reviewers: O1 Blessed Committers, valerio.bozzolan
Reviewed By: O1 Blessed Committers, valerio.bozzolan
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15414
Differential Revision: https://we.phorge.it/D25241
Summary:
`preg_match()` does not accept passing null as the `$subject` string parameter in PHP 8.1.
Thus add a `phutil_nonempty_string()` check if the `$subject` parameter is a non-empty string.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
```
EXCEPTION: (RuntimeException) preg_match(): Passing null to parameter #2 ($subject) of type string is deprecated at [<arcanist>/src/error/PhutilErrorHandler.php:261]
arcanist(head=arcpatch-D25250, ref.master=18554ea76ceb, ref.arcpatch-D25250=fade4603a799), phorge(head=diffusionEmptyBranchHistory, ref.master=e11c5486c92b, ref.diffusionEmptyBranchHistory=76f042e4b969)
#0 <#2> PhutilErrorHandler::handleError(integer, string, string, integer) called at [<arcanist>/src/error/PhutilErrorHandler.php:261]
#1 <#2> preg_match(string, NULL) called at [<phorge>/src/applications/diffusion/request/DiffusionGitRequest.php:6]
```
Closes T15435
Test Plan: After applying this change, try to "Clone" a Repository with no URI set shows the overlay dialog `Clone Repository - Repository has no URIs set.` as expected.
Reviewers: O1 Blessed Committers, valerio.bozzolan
Reviewed By: O1 Blessed Committers, valerio.bozzolan
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15435
Differential Revision: https://we.phorge.it/D25266
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
Closes T15420
Test Plan: Applied these two changes; afterwards "Browse Repositories" overlay dialog gets rendered in web browser and lists existing repositories.
Reviewers: O1 Blessed Committers, valerio.bozzolan
Reviewed By: O1 Blessed Committers, valerio.bozzolan
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15420
Differential Revision: https://we.phorge.it/D25244
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
Similarly, `preg_match()` does not accept passing `null` as `$subject` parameter since PHP 8.1.
Closes T15337
Test Plan: Applied these four changes and `/diffusion/4/manage/` finally rendered in web browser.
Reviewers: O1 Blessed Committers, valerio.bozzolan
Reviewed By: O1 Blessed Committers, valerio.bozzolan
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15337
Differential Revision: https://we.phorge.it/D25182
Summary:
`strlen()` was used in Phabricator to check if a generic value is a non-empty string.
This behavior is deprecated since PHP 8.1. Phorge adopts `phutil_nonempty_string()` as a replacement.
Note: this may highlight other absurd input values that might be worth correcting
instead of just ignoring. If phutil_nonempty_string() throws an exception in your
instance, report it to Phorge to evaluate and fix that specific corner case.
Closes T15300
Test Plan: Applied these three changes (on top of D25144, D25145, D25146, D25147) and `/diffusion/` finally rendered in web browser.
Reviewers: O1 Blessed Committers, avivey
Reviewed By: O1 Blessed Committers, avivey
Subscribers: speck, tobiaswiese, valerio.bozzolan, Matthew, Cigaryno
Maniphest Tasks: T15300
Differential Revision: https://we.phorge.it/D25150
Summary:
After this change, the page "/manage/" related to Diffusion
repositories has an "Actions" button for mobile users:
{F276113}
Before this change, that button was missing only from this
specific page, and so it was really difficult for them
(if not impossible) to edit repository's basic information:
{F276115}
Do not confuse the hamburger menu on the sidebar with this
"Actions". The hamburger contains queries like Active Repo,
All Repo, etc. Instead the Actions contains
Edit Basic Info, Allow Dangerous Changes, etc.
It seems the cause of this problem is originated from an
apparently incomplete rollback occurred in Sep 6 2017 from
Phabricator:
e1fd74ddb5
As you can see, the DiffusionRepositoryBasicsManagementPanel
was probably not completely restored from the previous
version, at least for the part related to the method
newActionList():
bb369c7b71
The method newActionList() was restored in every other
panel but no the one related to the Basics Management. So,
a new PhabricatorActionListView object was manually
created, but without generating an ID from the Celerity
utilities.
In short, this change assigns an ID to this Curtain, in
order to correctly generates a related "Actions" menu
again. Otherwise, on mobile the curtain (with all the
actions) is automatically hidden, without any alternative.
Closes T15067
Test Plan:
- Repository > Actions > Manage Repository
- Reduce the page until "Actions" appears (yeeh!)
Reviewers: O1 Blessed Committers, Cigaryno, avivey
Reviewed By: O1 Blessed Committers, Cigaryno, avivey
Subscribers: speck, tobiaswiese, Matthew, Cigaryno
Tags: #diffusion, #mobile
Maniphest Tasks: T15067
Differential Revision: https://we.phorge.it/D25109
Summary:
Note that PHP 8.2 implemented this deprecation:
Using ${var} in strings is deprecated, use {$var} instead
This change fixes some known cases.
Ref T15196
Test Plan: - I checked with my big eyes that everything is as it should be
Reviewers: O1 Blessed Committers, avivey
Reviewed By: O1 Blessed Committers, avivey
Subscribers: avivey, speck, tobiaswiese, Matthew, Cigaryno
Maniphest Tasks: T15196
Differential Revision: https://we.phorge.it/D25098
Summary: Ref T13682. Many subclasses of "CursorPagedPolicyAwareQuery" have the same implementation of "loadPage()", and this is a sensible default behavior.
Test Plan: Examined changes to verify that all removed methods have the same behavior.
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T13682
Differential Revision: https://secure.phabricator.com/D21838
Summary: Ref T13658
Test Plan:
This test plan is non-exhaustive.
- Looked at some of the config.
- Looked at guides.
Maniphest Tasks: T13658
Differential Revision: https://secure.phabricator.com/D21769
Summary:
Ref T13667. The Herald "content added" rule (and other similar rules) do not correctly extract content from empty commits.
When we load an empty raw diff, return an empty changed content map.
Ref T13588. Also fix some PHP8.1 null/string stuff
Test Plan:
- Ran "bin/repository reparse --publish <commit>", with an empty commit hash and a nonempty commit hash.
- Reviewed Herald transcripts for general sanity.
Maniphest Tasks: T13667, T13588
Differential Revision: https://secure.phabricator.com/D21761
Summary:
Ref T13673. After the changes in that task, we may execute "git config -l ..." as a user other than the user we used to write this temporary file.
Use "--file -" to pass the data instead, avoiding use of temporary files. This makes us agnostic to filesystem permissions.
Test Plan: Viewed a Git repository with submodules in Diffusion with "ssh.user" configured as a user relatively isolated from the webserver user.
Maniphest Tasks: T13673
Differential Revision: https://secure.phabricator.com/D21759
Summary:
See T13673. Recent versions of Git (and older versions with backported security patches) now refuse to run Git commands if the top-level repository directory is not owned by the user running the command.
Currently, we "sudo" to that user only when performing writes, so upgrading Git can aggressively break a Phabricator system by knocking out essentially all Diffusion/Conduit read pathways.
As an immediate mitigation, just "sudo" in all cases where a daemon user is available. This fixes the problem, and seems like the least-bad approach. The downside is that the web user may theoretically have fewer privileges than the daemon user and this could reduce the number of layers an attacker armed with some other Git vulnerability might have to get through to do something dangerous (e.g., perform a write on a pathway where only reads are expected), but any separation between the web and daemon accounts is essentially theoretical and has never been enforced.
Test Plan: Applied patch to impacted Phacility shard, saw Diffusion work properly again.
Differential Revision: https://secure.phabricator.com/D21756
Summary: Ref T13666. See T13666#256253. The order of these parameters is incorrect (introduced in D18817, which was trying to turn the revset "A or B" into "B or A"), but when "commit" is the single head of "branch" (which is common), the revset is functionally equivalent.
Test Plan:
An easy reproduction case is to make a "diffusion.historyquery" call manually against a Mercurial repository, with a valid "branch" name and some non-head "commit".
- Before patch, queried "head^" (by commit hash, not by symbol) of "default" and got "head" too (wrong).
- Applied patch to `secure001.phacility.net` (my local `hg` web environment is iffy right now).
- Re-ran the same query, saw "head^" as the first result rather than "head" (all hashes rather than symbols, as above), which is desired.
Reviewers: cspeckmim
Reviewed By: cspeckmim
Maniphest Tasks: T13666
Differential Revision: https://secure.phabricator.com/D21722
Summary:
The history query for the repository page isn't actually used to display any content. It looks like it was previously used to display the last user which modified a file however this looks to be removed in D21404. This removes the history query from happening as well as updates `DiffusionBrowseTableView` to remove the parameters for passing this information in, resulting in also updating `DiffusionBrowseController` to no longer need to put this information together.
Refs T13666
Test Plan:
1. I removed commits from a repository on the local state.
2. I navigated to the repository's landing page and saw that the landing page attempted to render content and only failed to load the browse files section.
3. I navigated to the history tab and verified that it showed an exception about failing to query commit information.
4. I restored the repository working state to function properly.
5. I navigated to a repository's landing page and verified it loaded properly, including showing the last modified date for each file.
6. I navigated to the Code, Branches, Tags, and History tabs to verify each tab page loaded properly.
7. I verified on the Code tab that the last modified date for each file displayed properly.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T13666
Differential Revision: https://secure.phabricator.com/D21717
Summary:
Refer to discussion on D21677#275541
Refs D21681 (arcanist changes)
Phabricator has several uses of the `--debug` flag being used with Mercurial. Use of this flag causes additional output to be added which Phabricator needs, however the behavior of `--debug` is not guaranteed to be stable, and in newer versions of Mercurial there have been additional output that has caused Phabricator to choke on parsing the output. This change removes several uses of `--debug` in favor of using `--template` with the `hg log` or `hg annotate` commands in combination with the `{p1.node}` or `{p2.node}` template format.
The use of `{p1node}` format in templates was added in [[ https://www.mercurial-scm.org/wiki/WhatsNew/Archive#Mercurial_2.4_.282012-11-01.29 | Mercurial 2.4 (2012) ]]. This format was deprecated in [[ https://www.mercurial-scm.org/wiki/WhatsNew#Mercurial_4.9_.282019-02-01.29 | Mercurial 4.9 (2019) ]] in favor of using `{p1.node}` format which is unclear when this new format was added (presumably earlier than Mercurial 4.9).
The use of `--template` with `hg annotate` is only officially supported in [[ https://www.mercurial-scm.org/wiki/Release4.6 | Mercurial 4.6 (2018) ]], though does appear to work in 4.5 but is not documented.
Since the `{p1node}` format was introduced in 2.4 this bumps the required version of `hg` to 2.4 (from 1.9). Since the `annotate --template` feature wasn't added until 4.6 (which is still fairly recent), the use of it is gated on a capability test, but still preferred for use where possible to avoid extraneous output from `--debug` flag.
Test Plan:
I verified I could do the following in a mercurial repository, while having mercurial 5.8 installed:
1. Navigate and view files in Diffusion under e.g. `/source/test-repo/`.
2. While viewing a file in Diffusion verified that I could view the blame of the file and the history/annotations looked accurate for the files I was browsing.
3. From the blame sidebar, select to view a commit which loaded and displayed changes properly.
4. View the history of the repository under e.g. `/source/test-repo/history/default/`. I verified the history looked correct and the tree-like structure showing relationship of commits also looked accurate.
I setup mercurial to run version 4.4, created a new repository, added some commits, and verified all the above behavior still works properly.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D21679
Summary:
With newer versions of Mercurial come newer debug messages which need filtered out.
1. In the scenario of Phabricator observing a hosted Mercurial repository which exists on a server in a multi-user environment it's possible that a repository computes branch cache at a tip revision which is not present. When this happens Mercurial will include in the debug output this information. This message indicates that the cache is going to be re-computed. See https://www.mercurial-scm.org/pipermail/mercurial/2014-June/047239.html.
2. Likely in some version with added or improved support for `pager` the debug info seems to indicate when a pager is being invoked for a command. This seems to print out regularly despite piping the stdout.
3. If the repository on Phabricator ever had the `largefiles` extension enabled then some additional details about "updated patterns" will print out.
Test Plan:
I verified an observed repository's history could be browsed, specifically the history of files which previously resulted in "Undefined offset: 1".
Added a unit test to check the results of `filterMercurialDebugOutput()`.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D21677
Summary:
Ref T13614. When a script holds the write lock but modifies the repository directly (rather than by pushing), the repository version won't change when the script releases the write lock. Thus, the writes may not propagate to other nodes (it depends which node lucks out and accepts the next write).
To guarantee that writes propagate, allow these scripts to pretend they pushed the repository. These are bare-bones valid events flagged as "Maintenance".
Test Plan:
- Wrote a script to hold the write lock, wait (or pretend to do something), then release the write lock.
- Applied patches, modified script to use new APIs ("newMaintenanceEvent()").
- Ran script, saw repository verison bump and relevant push logs:
{F8814923}
Maniphest Tasks: T13614
Differential Revision: https://secure.phabricator.com/D21670
Summary:
Ref T13614. When an omnipotent user calls "synchronizeWorkingCopyBeforeWrite()", we record a WorkingCopyVersion record with a null "userPHID". The UI then renders this as "Unknown Object (????)".
Improve this behavior:
- When no PHID is available, just render nothing in the UI (this doesn't seem meaningfully different from no version existing at all).
- Allow callers to provide an acting user PHID, similar to Editor.
There's currently no way to perform this kind of write legitimately in the upstream, but T13614 is providing one.
Test Plan:
- Wrote a script that calls "synchronizeWorkingCopyBeforeWrite()" as the omnipotent user.
- Ran script, saw "Unknown Object (????)" in the UI.
- Applied UI fix, saw empty UI.
- Applied "acting as" fix, modified script to act as the Diffusion application, ran script, saw "Diffusion" attribution in UI.
{F8814806}
Maniphest Tasks: T13614
Differential Revision: https://secure.phabricator.com/D21669
Summary:
Ref T13647. The ref discovery process prunes commits that no longer exist in the repository before executing "git log <new heads> --not <old heads>" to identify newly published commits.
If we don't do this, the "git log" command will fail if any old head has been pruned from the repository.
Currently, this test for missing commits starts with a call to "git for-each-ref" to attempt to resolve symbols as tag or branch names, but:
- this is painfully slow in repositories with many refs; and
- this is incorrect (not consistent with "git" behavior) for 40-character hex strings, which Git will never resolve as symbolic names.
Instead, when a symbol is a 40-character hex string, skip "git for-each-ref" and jump directly to "git cat-file --batch-check".
Test Plan:
- Ran `bin/repository update` in a repository with 65K refs and extra debugging info.
- Before: took ~30s, three calls to `git for-each-ref`.
- After: took ~20s, two calls to `git for-each-ref`. Same resolution result on queries.
Maniphest Tasks: T13647
Differential Revision: https://secure.phabricator.com/D21658
Summary:
Ref T13644. Ref T13638.
- Double-encode the symbol that is used as a path component, similar to Diffusion.
- Fix an outdated reference to ".path", which provided context for symbol lookup.
- Prevent command-clicking headers from looking up the path as a symbol.
Test Plan:
- Command-clicked headers, no longer got a symbol.
- Command-clicked stuff with "/", saw it double-encoded and decoded properly.
- Command-clicked normal symbols, saw "path" populate correctly.
Maniphest Tasks: T13644, T13638
Differential Revision: https://secure.phabricator.com/D21641
Summary: Ref T13641. Make "active bindings" a real query and make callers that only care about active bindings only query for active bindings.
Test Plan:
- Queried for "bindings" and "activeBindings" via Conduit.
- Disabled/enabled devices, saw binding status update in UI.
- Loaded Diffusion cluster layout.
- Grepped for `needBindings()`, `getActiveBindings()`, etc.
Subscribers: yelirekim, PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T13641
Differential Revision: https://secure.phabricator.com/D21628
Summary: Ref T13639. Move Diffusion to use the new API and get rid of the old API now that it no longer has any callers.
Test Plan:
Grepped for remaining callers.
{F8539335}
Maniphest Tasks: T13639
Differential Revision: https://secure.phabricator.com/D21620
