Alias system

[NicolasMahe]

Fix #558 about re-using the same volumes.
Closes #607, closes #605 and closes #612

This PR introduce a new optional property called alias in mesg.yml.
The alias can be use instead of the generated service id in any API and command that accept a service ID.
If the alias is omitted, one is generated based on the service hash.
If the alias already exist, it get replace by the last deployed service.

There is no breaking change.

[krhubert]

It would be great if we can provide versioning along with alias.

[ilgooz]

It would be great if we can provide versioning along with alias.

What do you mean exactly? We still have the id of service which is its version. Do you've something like semver in your mind? Semver is much more readable than what we have now. Like docker containers we can have these short names: http-server:latest, http-server:3.8. But having long hash IDs might be required while running on blockchain to easily identify the author of a service by service's source code. I'm not sure, maybe @NicolasMahe & @antho1404 can share their opinions about this.

[antho1404]

The idea with the versioning is to have a list of hashes for a specific alias instead of a single hash for a specific alias with by default the first one (or last one) is the latest deployment but we could go back in time with all the deployments and see for the same alias all the different versions updated.

It's not really versions like semver, it's more all the previous versions deployed. (Is it what you were thinking @krhubert ?)

This is something we might do later on but i really wanted to keep it simple for now

[mesg-bot]

This pull request has been mentioned on MESG Community. There might be relevant details there:

https://forum.mesg.com/t/fixed-service-id-name-service-update-create-feature/110/4

[NicolasMahe]

@ilgooz > It's a bit better to do atomic reads (with transactions) for Get & All as well.

Transaction are only useful to group multiple reads and writes into a single atomic transaction.
If it only writes, batch is enough.
If it only one read, no need for transaction.
If it multiple reads, transactions are useful.

For the All function, i don't think we need transaction because Iterator is returning a snapshot of the database. So concurrent modifications are not affecting the iterator.

[NicolasMahe]

I add this test in order to make sure that Delete function behave correctly on concurrent deletion.
If you remove the transaction in Delete function, then this test fails.

To test, replace in Delete function tx, err := d.db.OpenTransaction() by tx = db.db and remove tx.Discard() and tx.Commit()

[NicolasMahe]

@mesg-foundation/core please review. I add few tests and change some minor code.
Please don't commit anything on this PR

[ilgooz]

@ilgooz > It's a bit better to do atomic reads (with transactions) for Get & All as well.

Transaction are only useful to group multiple reads and writes into a single atomic transaction.
If it only writes, batch is enough.
If it only one read, no need for transaction.
If it multiple reads, transactions are useful.

For the All function, i don't think we need transaction because Iterator is returning a snapshot of the database. So concurrent modifications are not affecting the iterator.

Here there is multiple reads https://github.com/mesg-foundation/core/blob/64407a543a353286229e8ba76d23f264d2abeeb4/database/service_db.go#L127

[NicolasMahe]

@ilgooz > It's a bit better to do atomic reads (with transactions) for Get & All as well.
Transaction are only useful to group multiple reads and writes into a single atomic transaction.
If it only writes, batch is enough.
If it only one read, no need for transaction.
If it multiple reads, transactions are useful.
For the All function, i don't think we need transaction because Iterator is returning a snapshot of the database. So concurrent modifications are not affecting the iterator.

Here there is multiple reads

https://github.com/mesg-foundation/core/blob/64407a543a353286229e8ba76d23f264d2abeeb4/database/service_db.go#L127

Nice catch. Done in c486810aee9d9f325e136a3f98625c5d50615753

[ilgooz]

https://github.com/mesg-foundation/core/pull/583/files#r238167227

resolved