Summary:
Ref T2787. Currently, we kill a cart and dead-end the workflow on a charge failure.
Instead, fail the charge and reset the cart so the user can try using a valid payment instrument like a normal checkout workflow would.
Some shakiness/smoothing on WePay for the moment; PayPal is still made up since we don't have a "Hold" state yet.
Test Plan: {F215214}
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10666
Summary: Fixes T4018. Basically hits the bullet points in that task description except the "ideally" one.
Test Plan:
ran bin/config migrate and saw sensible output.
```
~> ./bin/config migrate
Migrating file-based config to more modern config...
Skipping config of source type PhabricatorConfigDatabaseSource...
Skipping config of source type PhabricatorConfigLocalSource...
Skipping config of source type PhabricatorConfigDefaultSource...
Done. Migrated 0 keys.
```
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: hach-que, epriestley, Korvin
Maniphest Tasks: T4018
Differential Revision: https://secure.phabricator.com/D10490
Summary:
Ref T2787. This has some rough edges but basically works.
- Users can cancel orders that are in incomplete states (or in complete states, if the application allows them to -- for example, some future application might allow cancellation of billed-but-not-shipped orders).
- Merchant controllers can partially or fully refund orders from any state after payment.
Test Plan: This is still rough around the edges, but issued Stripe and WePay refunds.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: chad, epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10664
Summary:
Ref T2787.
- Allow merchants to disable payment providers.
- Show more useful information about providers on the payments page.
- Make test vs live more clear.
- Show merchant status.
- Add a description to merchants to flesh them out a bit -- the merchant areas of responsibilities seem to be fitting well with accounts, etc.
Test Plan: {F215109}
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10662
Summary: Fixes T6265, allows you to pass required:false as a parameter.
Test Plan: Add required:false to a field, no longer see "Required"
Reviewers: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6265
Differential Revision: https://secure.phabricator.com/D10659
Summary: Ref T2787. Uses the real icons. Straightens out the add payment flow a tiny bit.
Test Plan: {F214922}
Reviewers: btrahan, chad
Reviewed By: chad
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10654
Summary: Fixes T6252
Test Plan: Test project query from conduit app, see no errors in log.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6252
Differential Revision: https://secure.phabricator.com/D10655
Summary: Ref T6256, this prevents more installs from getting in this weird state. We'll have to follow up if possible to "fix" the issue retroactively.
Test Plan: Test moving a backlog column to new position, hiding rest of other panels.
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6256
Differential Revision: https://secure.phabricator.com/D10651
Summary:
Ref T2787. Builds on D10649 by rebining existing objects (carts, charges, etc) to merchantPHIDs and providerPHIDs instead of an implicit global merchant and weird global artifacts (providerType / providerKey).
Basically:
- When you create something that users can pay for, you specify a merchant to control where the payment goes.
- Accounts are install-wide, but payment methods are bound to merchants. This seems to do a reasonable job of balancing usability and technical concerns.
- Replace a bunch of weird links between objects with standard PHIDs.
- Improve "add payment method" flow.
Test Plan: Went through the Fund flow with Stripe and WePay, funding an initiative.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10652
Summary:
Ref T2787. Instead of making providers global configuration, make them a thing on merchants with web configuration.
Payment methods and some of the pyament workflow needs to be retooled a bit after this, but this seemed like a reasonable cutoff point for this diff.
Test Plan: See screenshots.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10649
Summary:
Ref T2787. Currently, you add payment providers (Stripe, Paypal, etc) in global configuration.
Generally, this approach is cumbersome, limiting, and often hard for users to figure out. It also doesn't provide a natural way to segment payment receivers or provide web access to administrative payment functions like issuing refunds, canceling orders, etc. I think that stuff definitely needs to be in the web UI, and the rule for access to it can't reasonably just be "all administrators" in a lot of reasonable cases.
The only real advantage is that it prevents an attacker from adjusting settings and pointing something at an account they control. But this attack can be mitigated through notifications, some sort of CLI-only merchant lock, payment accounts being relatively identifiable, etc.
So introduce "merchants", which are basically payable entities. An individual merchant will have attached Paypal, Stripe, etc., accounts, and access rules. When you buy something in an application, the merchant to pay is also specified. They also provide an umbrella for dealing with permissions down the line.
This may get a //little// cumbersome because if there are several merchants your saved card information is not shared across them. I think that will be fine in the normal case (most installs will have only one merchant). Even if it isn't and we leave providers global, I think introducing this is the right call from a web UI / permissions point of view. I'll play around with it in the next couple of diffs and figure out exactly where the line goes.
Test Plan: Listed, created, edited, viewed merchants.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10648
Summary:
Ref T2787. These were still stuck in the stone ages.
(The handles are pretty skeletal but most aren't used anywehre.)
Test Plan: Funded an initiative without anything breaking. Grepped for removed constants.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10647
Summary: Ref T2787. Like Stripe, this one is pretty easy to get working correctly on the "good" path and fataling out in a safe way on bad paths.
Test Plan: Funded an initiative with Balanced.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10645
Summary:
Ref T2787. For test charges, Paypal is putting the charge in a "payment review" state. Dealing with this state requires way more infrastructure than other providers: we're supposed to pause delivery, then poll Paypal every 6 hours to see if the review has resolved.
Since I can't seem to generate normal test charges, I can't test Paypal for now. Disable it until we have more infrastructure.
(This diff gets us further along, up to the point where I hit this issue.)
Test Plan: Read documentation, rolled eyes.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10644
Summary:
Ref T2787. This basically already works correctly since the hard logic is external to the provider on API providers. Tweak a couple of things.
Failures still just fail the cart completely, for now.
Test Plan: Completed a charge with Stripe.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10640
Summary:
Ref T2787. This doesn't get all the edge cases quite correct, but is generally a safe, complete payment workflow:
- Shares the actual charging state logic.
- Makes it appropriately stateful with locking and transactions.
- Gets the main flow correct.
- Detects failure cases, just tends to blow up rather than help the user resolve them.
Test Plan:
- Charged with WePay.
- Charged with Infinite Free Money.
- Resumed an abandoned cart.
- Hit all failure states where we just dead-end the cart. Not ideal, but (seemingly) complete/safe/correct.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10639
Summary: Ref T2787. Similar to D10634, give applications more control over the cart workflow. For now this just means they get to pick exit URIs, but in the future they can manage more details of cart behavior.
Test Plan: Funded an initiative and got returned to the initiative instead of dead-ending in Phortune.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10638
Summary: Fixes T6244, adds icons for payment providers. May split into different sprites down the road.
Test Plan: Photoshop
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6244
Differential Revision: https://secure.phabricator.com/D10642
Summary: just explicitly check if the file doesn't exist *first*, and then do the standard include thing with the more generic error if that doesn't work. Fixes T6255.
Test Plan: re-started apache and phabricator still worked; will ask csilvers to give it a whirl too
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6255
Differential Revision: https://secure.phabricator.com/D10643
Summary: without escapage here, creating databases fails. Fixes T6251.
Test Plan: ran the command CREATE DATABASE foo COLLATION binary and it failed; ran the command CREATE DATABASE foo2 COLLATION "binary" and it worked; trusting that the %T still works as advertised.
Reviewers: chad, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6251
Differential Revision: https://secure.phabricator.com/D10641
Summary: Fixes T6254 and renames status as string. Though maybe this should go through `formatStringConstants`?
Test Plan: Reload Conduit page, see new text.
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6254
Differential Revision: https://secure.phabricator.com/D10637
Summary:
Ref T2787. When a user purchases a product in Phortune, transition the cart through a purchased state and invoke product callbacks so applications can respond to the workflow.
Also shore up some stuff like preventing negative amounts of funding.
Test Plan: Backed an initiative and saw it show up on the initiative after completing the purcahsing workflow.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10635
Summary: Ref T2787. `Product` is currently a fairly heavy object, but as Phortune develops it makes a lot of sense to make it a lighter object and put more product logic in applications. Convert it into a fairly lightweight reference to applications. The idea is that Phortune is mostly providing a cart flow, and applications manage the details of products.
Test Plan: Funded an initiative for $1.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10634
Summary:
Ref T2787. Phortune currently stores a bunch of stuff as `...inUSDCents`. This ends up being pretty cumbersome and I worry it will create a huge headache down the road (and possibly not that far off if we do Coinbase/Bitcoin soon). Even now, it's more of a pain than I figured it would be.
Instead:
- Provide an application-level serialization mechanism.
- Provide currency serialization.
- Store currency in an abstract way (currently, as "1.23 USD") that can handle currencies in the future.
- Change all `...inUSDCents` to `..asCurrency`.
- This generally simplifies all the application code.
- Also remove some columns which don't make sense or don't make sense anymore. Notably, `Product` is going to get more abstract and mostly be provided by applications.
Test Plan:
- Created a new product.
- Purchased a product.
- Backed an initiative.
- Ran unit tests.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2787
Differential Revision: https://secure.phabricator.com/D10633
Summary: Ref T1191. We don't create new databases with appropriate collation yet.
Test Plan:
Created a new database and saw it issue:
```
>>> [10] <query> CREATE DATABASE IF NOT EXISTS `phabricator2_testo` COLLATE utf8mb4_bin
```
Reviewers: btrahan, hach-que
Reviewed By: hach-que
Subscribers: epriestley
Maniphest Tasks: T1191
Differential Revision: https://secure.phabricator.com/D10632
Summary:
Ref T4209. This creates storage for public keys against authorized hosts, such that servers can be authorized to make Conduit calls as the omnipotent user.
Servers are registered into this system by running the following command once:
```
bin/almanac register
```
NOTE: This doesn't implement authorization between servers, just the storage of public keys.
Placing this against Almanac seemed like the most sensible place, since I'm imagining in future that the `register` command will accept more information (like the hostname of the server so it can be found in the service directory).
Test Plan: Ran `bin/almanac register` and saw the host (and public key information) appear in the database.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Maniphest Tasks: T4209
Differential Revision: https://secure.phabricator.com/D10400
Summary: Fixes T6119. This is a little fuzzy, but generally bumping up `innodb_buffer_pool_size` to something bigger than the default (which is often anemic, at `8M`) is desriable, and it seems like it will fix the specific issue a user encountered in T6119.
Test Plan: {F211855}
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6119
Differential Revision: https://secure.phabricator.com/D10630
Summary:
Fixes T6230. These files have not been read by default for a long time, but users are frequently confused and try to edit `default.conf.php`.
Remove the actual files. Allow `phabricator_read_config_file(...)` to continue working as though they exist so as to not break config-file-based installs.
Test Plan:
I used this script to make sure that removing `default.conf.php` won't change things for installs which are still using config files:
```
<?php
require_once 'scripts/__init_script__.php';
$file = require 'conf/default.conf.php';
$global = new PhabricatorConfigDefaultSource();
$global_values = $global->getAllKeys();
foreach ($file as $key => $value) {
$global_value = idx($global_values, $key, (object)array());
if ($value !== $global_value) {
echo "{$key}\n\n";
echo "FILE VALUE\n";
var_dump($value);
echo "\n";
echo "DEFAULT VALUE\n";
var_dump($global_value);
return;
}
}
```
These were the keys that had issues:
- `log.access.format` Not specified in default.conf.php, safe to speciy.
- `mysql.pass` Empty string in file, null in global. Same effect.
- `metamta.default-addrress` One used `noreply@example.com`, one `noreply@phabricator.example.com`. These are just human-readable examples so it's safe to change behavior.
- `metamta.domain` same as above, `example.com` vs `phabricator.example.com`.
- `phpmailer.smtp-host` One used null, one empty string.
- `phpmailer.smtp-protocol` As above.
- `files.viewable-mime-types` File version is out of date.
- `repository.default-local-path` Null in file, set in global. This is correct to set to a default value now.
- `pygments.dropdown-choices` File version is out of date.
- `environment.append-paths` File version is empty, global version adds common paths. This //could// change behavior, but the web behavior is better and more reasonable in general, and a system would need to be configured in a very bizarre way for this to be relevant.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6230
Differential Revision: https://secure.phabricator.com/D10628
Summary:
Ref T1191. Although I fixed some of the mutations earlier (in D10598), I missed the column mutations under old versions of MySQL. In particular, this isn't valid:
- `ALTER TABLE ... MODIFY columnName VARCHAR(64) COLLATE binary`
Issue the permitted version of this instead, which is:
- `ALTER TABLE ... MODIFY columnName VARBINARY(64)`
Also fixed an issue where a clean schema had the wrong nullability for a column in the draft table. Force it to the expected nullability.
The other trick here is around the one column with a FULLTEXT index on it, which needs a little massaging.
Test Plan:
- Forced my local install to return `false` for utf8mb4 support.
- Did a clean adjust into `binary` columns.
- Poked around, added emoji to things.
- Reverted the fake check and did a clean adjust into `utf8mb4` columns.
- Emoji survived.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: fabe, epriestley
Maniphest Tasks: T1191
Differential Revision: https://secure.phabricator.com/D10627
Summary: thanks mailbox
Test Plan: unit tests
Reviewers: btrahan, chad
Reviewed By: chad
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D10629
Summary:
Ref T1191. After utf8mb4 conversion, these tests no longer pass because MySQL allows emoji and gclefs and such.
We could keep these tests running by keeping a `ut8f_bin` table around somewhere, but we have no other use cases for it and it does not seem worth the added complexity. All these BMP-only codepaths are on the way out.
Update the `%s` / `%B` test to make sure it's rejecting invalid byte sequences, which are still not permitted.
Test Plan: Tests now pass.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T1191
Differential Revision: https://secure.phabricator.com/D10621
Summary:
Ref T1191. Now that the whole database is covered, we don't need to do as much work to build expected schemata. Doing them database-by-database was helpful in converting, but is just reudndant work now.
Instead of requiring every application to build its Lisk objects, just build all Lisk objects.
I removed `harbormaster.lisk_counter` because it is unused.
It would be nice to autogenerate edge schemata, too, but that's a little trickier.
Test Plan: Database setup issues are all green.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley, hach-que
Maniphest Tasks: T1191
Differential Revision: https://secure.phabricator.com/D10620
Summary: Ref T1191. The index's case sensitivity depends on the column type. Using `text` makes the search case-sensitive, which is not desirable.
Test Plan: After adjustment, searched for "PROJECTS" and found hits against "projects".
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T1191
Differential Revision: https://secure.phabricator.com/D10619
Summary: Fixes T6211. This gives Herald rules an explicit execution order, which seems generally good. See some discussion on T6211 and inline.
Test Plan:
- Added unit test.
- Dry ran rules and saw rules appear in the expected order in the transcript.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6211
Differential Revision: https://secure.phabricator.com/D10624
Summary: Fixes T6210. The current messaging may be confusing if `pygmentize` is available but broken.
Test Plan: Faked the binary names and hit the errors, which seemed helpful.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6210
Differential Revision: https://secure.phabricator.com/D10626
Summary: Ref T6201. This isn't quite perfect but should be good enough. At some point far in the future I plan to revamp feed rendering a bit. This should possibly become a real ApplicationTransaction story eventually, too.
Test Plan: {F211777}
Reviewers: btrahan, chad
Reviewed By: chad
Subscribers: epriestley
Maniphest Tasks: T6201
Differential Revision: https://secure.phabricator.com/D10625
Summary:
Ref T6223. Two issues:
- We don't use `/u` mode on these regexps. Without `/u`, the `\w`/`\W`/`\s`/`\S` modifiers have bad behavior on non-ASCII bytes. Add the flag to use unicode mode, making `\w` and `\s` behave like we expect.
- We might possibly want to do something different here eventually (for example, if the `/u` flag has some huge performance penalty) but this seems OK for now.
- We use `\b` (word boundary) to terminate the match, but `🐳` is not a word character. Use `(?!\w)` instead ("don't match before a word character") which is what we mean.
Test Plan: {F211498}
Reviewers: btrahan, chad
Reviewed By: chad
Subscribers: epriestley
Maniphest Tasks: T6223
Differential Revision: https://secure.phabricator.com/D10618
Summary: Ref T1191. The bulk of the slowness in T1191 is copying tables. In some cases, we can't avoid this, but we have various readthrough caches which may be very large and are safe to drop, and dropping them is very quick (much less than 1 second). In particular, dropping the `changeset_parse_cache` made the process at least ~8 minutes faster on `secure.phabricator.com` (I killed it after 8 minutes, so I'm not sure what the real number is).
Test Plan: Ran `bin/storage adjust` and saw it drop caches before applying adjustments.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T1191
Differential Revision: https://secure.phabricator.com/D10616
Summary:
Ref T1191. Long ago, Maniphest generated with 40-character mail keys. These prevent the migration to `bytes20`. We had about 300 of these on secure.phabricator.com from several years ago.
Just truncate them. This adjusts reply-to addresses, but it's very likely that none are relevant anymore.
Test Plan: Ran migration on `secure.phabricator.com` to truncate keys.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T1191
Differential Revision: https://secure.phabricator.com/D10615
Summary: Ref T1191. This predates the mdoern patch stuff and may exist on very, very old installs. By the time they apply this patch, it's guaranteed it won't matter anymore. Drop it to make the schemata consistent with expectations.
Test Plan: Ran patch on installs with and without the table.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T1191
Differential Revision: https://secure.phabricator.com/D10611
Summary: Ref T1191. Similar issue to D10613. This column usually has a hash exactly 12 bytes long, but sometimes stores an internal builtin query name like "open", "all", etc. It might be nice to promote those to 12-byte hashes of a consistent length eventually, but for now just make this a variable-length column.
Test Plan: Ran migration, no longer saw issues with reordering builtin saved searches.
Reviewers: btrahan
Subscribers: epriestley
Maniphest Tasks: T1191
Differential Revision: https://secure.phabricator.com/D10614
Summary:
Ref T1191. The `bytes` types are BINARY(...), which is fixed-length and zero-pads. These hashes are not 64 characters long, so migrating them to `binary` ends up with a bunch of zero-padding.
Instead, migrate them to `text` so we drop the zero padding. It would be vaguely nice to either introduce a `varbytes` type (ick) or change the hash size to a standard size (nicer) eventually, but this isn't very important.
Test Plan: Will adjust `secure.phabricator.com`.
Reviewers: btrahan
Subscribers: epriestley
Maniphest Tasks: T1191
Differential Revision: https://secure.phabricator.com/D10613
