[API-321] Migrate tests for blast writes

[schottra]

This is a followup to #338 to migrate the quite sizable blasts test file. Source is here: https://github.com/AudiusProject/audius-protocol/blob/ab73d979f98e151ac0aa408c7a82c105c3b6355e/comms/discovery/rpcz/chat_blast_test.go#L760

• Migrated everything to pgx patterns
• Split the file into multiple isolated tests instead of one huge test
• Using fixtures instead of manual inserts for setup data
• Removed a lot of the 'in the past' time variables in favor of just using time.Now() for operations that are sequential
• Updated the version of getNewBlasts in the comms directory to match the one we use in the get endpoint. We are being intentionally WET here.
• Fixed an issue with upgrading blasts to chats where the timestamp that we assign to the new message would be in the wrong timezone if run on a machine not set to UTC. In practice, this won't break on CI or production, but it does break when running on a local machine 😢

As a bonus, also cleaned up the tests from the previous PR to match the patterns here.

[schottra]

Note here the usage of .UTC(). On my machine, when running this, the location() on this time.Time was being set to Local and that was causing us to insert messages in the past. Since chat_blast uses a timestamptz column type and chat_message uses a plain timestamp column type, pgx semantics when inserting will cause it to try and map from the timezone in the incoming variable to the timezone of the machine running postgres.
Kind of gnarly 😬

[raymondjacobson]

why remove?

[schottra]

I didn't think writing them would add a ton of value for the amount of time it would take. Can be persuaded though :-p

[schottra]

This is a followup to #338 to migrate the quite sizable blasts test file. Source is here: https://github.com/AudiusProject/audius-protocol/blob/ab73d979f98e151ac0aa408c7a82c105c3b6355e/comms/discovery/rpcz/chat_blast_test.go#L760

• Migrated everything to pgx patterns
• Split the file into multiple isolated tests instead of one huge test
• Using fixtures instead of manual inserts for setup data
• Removed a lot of the 'in the past' time variables in favor of just using time.Now() for operations that are sequential
• Updated the version of getNewBlasts in the comms directory to match the one we use in the get endpoint. We are being intentionally WET here.
• Fixed an issue with upgrading blasts to chats where the timestamp that we assign to the new message would be in the wrong timezone if run on a machine not set to UTC. In practice, this won't break on CI or production, but it does break when running on a local machine 😢

As a bonus, also cleaned up the tests from the previous PR to match the patterns here.

[schottra]

Note here the usage of .UTC(). On my machine, when running this, the location() on this time.Time was being set to Local and that was causing us to insert messages in the past. Since chat_blast uses a timestamptz column type and chat_message uses a plain timestamp column type, pgx semantics when inserting will cause it to try and map from the timezone in the incoming variable to the timezone of the machine running postgres.
Kind of gnarly 😬

[raymondjacobson]

why remove?

[schottra]

I didn't think writing them would add a ton of value for the amount of time it would take. Can be persuaded though :-p

[schottra]

This is a followup to #338 to migrate the quite sizable blasts test file. Source is here: https://github.com/AudiusProject/audius-protocol/blob/ab73d979f98e151ac0aa408c7a82c105c3b6355e/comms/discovery/rpcz/chat_blast_test.go#L760

• Migrated everything to pgx patterns
• Split the file into multiple isolated tests instead of one huge test
• Using fixtures instead of manual inserts for setup data
• Removed a lot of the 'in the past' time variables in favor of just using time.Now() for operations that are sequential
• Updated the version of getNewBlasts in the comms directory to match the one we use in the get endpoint. We are being intentionally WET here.
• Fixed an issue with upgrading blasts to chats where the timestamp that we assign to the new message would be in the wrong timezone if run on a machine not set to UTC. In practice, this won't break on CI or production, but it does break when running on a local machine 😢

As a bonus, also cleaned up the tests from the previous PR to match the patterns here.

[schottra]

Note here the usage of .UTC(). On my machine, when running this, the location() on this time.Time was being set to Local and that was causing us to insert messages in the past. Since chat_blast uses a timestamptz column type and chat_message uses a plain timestamp column type, pgx semantics when inserting will cause it to try and map from the timezone in the incoming variable to the timezone of the machine running postgres.
Kind of gnarly 😬

[raymondjacobson]

why remove?

[schottra]

I didn't think writing them would add a ton of value for the amount of time it would take. Can be persuaded though :-p

[schottra]

This is a followup to #338 to migrate the quite sizable blasts test file. Source is here: https://github.com/AudiusProject/audius-protocol/blob/ab73d979f98e151ac0aa408c7a82c105c3b6355e/comms/discovery/rpcz/chat_blast_test.go#L760

• Migrated everything to pgx patterns
• Split the file into multiple isolated tests instead of one huge test
• Using fixtures instead of manual inserts for setup data
• Removed a lot of the 'in the past' time variables in favor of just using time.Now() for operations that are sequential
• Updated the version of getNewBlasts in the comms directory to match the one we use in the get endpoint. We are being intentionally WET here.
• Fixed an issue with upgrading blasts to chats where the timestamp that we assign to the new message would be in the wrong timezone if run on a machine not set to UTC. In practice, this won't break on CI or production, but it does break when running on a local machine 😢

As a bonus, also cleaned up the tests from the previous PR to match the patterns here.

[schottra]

Note here the usage of .UTC(). On my machine, when running this, the location() on this time.Time was being set to Local and that was causing us to insert messages in the past. Since chat_blast uses a timestamptz column type and chat_message uses a plain timestamp column type, pgx semantics when inserting will cause it to try and map from the timezone in the incoming variable to the timezone of the machine running postgres.
Kind of gnarly 😬

[raymondjacobson]

why remove?

[schottra]

I didn't think writing them would add a ton of value for the amount of time it would take. Can be persuaded though :-p

[schottra]

This is a followup to #338 to migrate the quite sizable blasts test file. Source is here: https://github.com/AudiusProject/audius-protocol/blob/ab73d979f98e151ac0aa408c7a82c105c3b6355e/comms/discovery/rpcz/chat_blast_test.go#L760

• Migrated everything to pgx patterns
• Split the file into multiple isolated tests instead of one huge test
• Using fixtures instead of manual inserts for setup data
• Removed a lot of the 'in the past' time variables in favor of just using time.Now() for operations that are sequential
• Updated the version of getNewBlasts in the comms directory to match the one we use in the get endpoint. We are being intentionally WET here.
• Fixed an issue with upgrading blasts to chats where the timestamp that we assign to the new message would be in the wrong timezone if run on a machine not set to UTC. In practice, this won't break on CI or production, but it does break when running on a local machine 😢

As a bonus, also cleaned up the tests from the previous PR to match the patterns here.

[schottra]

Note here the usage of .UTC(). On my machine, when running this, the location() on this time.Time was being set to Local and that was causing us to insert messages in the past. Since chat_blast uses a timestamptz column type and chat_message uses a plain timestamp column type, pgx semantics when inserting will cause it to try and map from the timezone in the incoming variable to the timezone of the machine running postgres.
Kind of gnarly 😬

[raymondjacobson]

why remove?

[schottra]

I didn't think writing them would add a ton of value for the amount of time it would take. Can be persuaded though :-p

[schottra]

This is a followup to #338 to migrate the quite sizable blasts test file. Source is here: https://github.com/AudiusProject/audius-protocol/blob/ab73d979f98e151ac0aa408c7a82c105c3b6355e/comms/discovery/rpcz/chat_blast_test.go#L760

• Migrated everything to pgx patterns
• Split the file into multiple isolated tests instead of one huge test
• Using fixtures instead of manual inserts for setup data
• Removed a lot of the 'in the past' time variables in favor of just using time.Now() for operations that are sequential
• Updated the version of getNewBlasts in the comms directory to match the one we use in the get endpoint. We are being intentionally WET here.
• Fixed an issue with upgrading blasts to chats where the timestamp that we assign to the new message would be in the wrong timezone if run on a machine not set to UTC. In practice, this won't break on CI or production, but it does break when running on a local machine 😢

As a bonus, also cleaned up the tests from the previous PR to match the patterns here.

[schottra]

Note here the usage of .UTC(). On my machine, when running this, the location() on this time.Time was being set to Local and that was causing us to insert messages in the past. Since chat_blast uses a timestamptz column type and chat_message uses a plain timestamp column type, pgx semantics when inserting will cause it to try and map from the timezone in the incoming variable to the timezone of the machine running postgres.
Kind of gnarly 😬

[raymondjacobson]

why remove?

[schottra]

I didn't think writing them would add a ton of value for the amount of time it would take. Can be persuaded though :-p

[schottra]

This is a followup to #338 to migrate the quite sizable blasts test file. Source is here: https://github.com/AudiusProject/audius-protocol/blob/ab73d979f98e151ac0aa408c7a82c105c3b6355e/comms/discovery/rpcz/chat_blast_test.go#L760

• Migrated everything to pgx patterns
• Split the file into multiple isolated tests instead of one huge test
• Using fixtures instead of manual inserts for setup data
• Removed a lot of the 'in the past' time variables in favor of just using time.Now() for operations that are sequential
• Updated the version of getNewBlasts in the comms directory to match the one we use in the get endpoint. We are being intentionally WET here.
• Fixed an issue with upgrading blasts to chats where the timestamp that we assign to the new message would be in the wrong timezone if run on a machine not set to UTC. In practice, this won't break on CI or production, but it does break when running on a local machine 😢

As a bonus, also cleaned up the tests from the previous PR to match the patterns here.

[schottra]

Note here the usage of .UTC(). On my machine, when running this, the location() on this time.Time was being set to Local and that was causing us to insert messages in the past. Since chat_blast uses a timestamptz column type and chat_message uses a plain timestamp column type, pgx semantics when inserting will cause it to try and map from the timezone in the incoming variable to the timezone of the machine running postgres.
Kind of gnarly 😬

[raymondjacobson]

why remove?

[schottra]

I didn't think writing them would add a ton of value for the amount of time it would take. Can be persuaded though :-p

[schottra]

This is a followup to #338 to migrate the quite sizable blasts test file. Source is here: https://github.com/AudiusProject/audius-protocol/blob/ab73d979f98e151ac0aa408c7a82c105c3b6355e/comms/discovery/rpcz/chat_blast_test.go#L760

• Migrated everything to pgx patterns
• Split the file into multiple isolated tests instead of one huge test
• Using fixtures instead of manual inserts for setup data
• Removed a lot of the 'in the past' time variables in favor of just using time.Now() for operations that are sequential
• Updated the version of getNewBlasts in the comms directory to match the one we use in the get endpoint. We are being intentionally WET here.
• Fixed an issue with upgrading blasts to chats where the timestamp that we assign to the new message would be in the wrong timezone if run on a machine not set to UTC. In practice, this won't break on CI or production, but it does break when running on a local machine 😢

As a bonus, also cleaned up the tests from the previous PR to match the patterns here.

[schottra]

Note here the usage of .UTC(). On my machine, when running this, the location() on this time.Time was being set to Local and that was causing us to insert messages in the past. Since chat_blast uses a timestamptz column type and chat_message uses a plain timestamp column type, pgx semantics when inserting will cause it to try and map from the timezone in the incoming variable to the timezone of the machine running postgres.
Kind of gnarly 😬

[raymondjacobson]

why remove?

[schottra]

I didn't think writing them would add a ton of value for the amount of time it would take. Can be persuaded though :-p