draft: Demo for aries vcx

[Patrik-Stas]

Partially addressing #870

• Added Alice & Faber demo to establish connection
• Modified simple_message_relay to enable its consumption also as library, rather than executable only
• Extended simple_message_relay with mpsc channels which forward any incoming message to mpsc receiver. This is nice because in the demo, we don't need to poll HTTP API for new messages - but rather instantly get notified in-memory.

Having to interact with aries-vcx from scratch from new application, some improvements I've identified:

• Oob builder for messages crate #949
• Minimize imports of agency_client #951

Note

This can be thought of as temporary solution to demonstrate the APIs and usage. And while far from ideal yet (todos left, issues created), it's a lot easier to understand than pointing someone to integration tests today. Main issue of integration tests is that they are all couploed with vcxagency-node mediator, but also other issues.
After integration tests are in better shape after:

• Declutter issuance&presentation protocols #945
• Remove msg-sending IO for issuer and holder #946
• and subsequent removal of IO, and therefore coupling with vcxagency-node mediator
• and some refactoring of "Alice" "Faber" devsetup agents

I believe that integration tests could be in condition to demonstrate the usage (and surface outstanding API issues) equally good as this demo.
But for the time being, this demo is a lot simpler than the existing integration tests.

Demo

Running the demo produces following logs

[codecov-commenter]

Codecov Report

Merging #952 (ab545cc) into main (4e65aec) will increase coverage by 30.32%.
The diff coverage is 0.00%.

❗ Current head ab545cc differs from pull request most recent head 58f03a7. Consider uploading reports for the commit 58f03a7 to get more accurate results

@@            Coverage Diff             @@
##            main     #952       +/-   ##
==========================================
+ Coverage   0.05%   30.38%   +30.32%     
==========================================
  Files        384      422       +38     
  Lines      21249    26230     +4981     
  Branches    3835     5084     +1249     
==========================================
+ Hits          12     7970     +7958     
+ Misses     21236    16041     -5195     
- Partials       1     2219     +2218     

Flag	Coverage Δ
unittests-aries-vcx	30.38% <0.00%> (+30.32%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files	Coverage Δ
aries_vcx/src/protocols/connection/generic/mod.rs	9.16% <ø> (+9.16%)	⬆️
aries_vcx/src/protocols/connection/invitee/mod.rs	78.94% <ø> (+78.94%)	⬆️
aries_vcx/src/protocols/connection/mod.rs	7.89% <ø> (+7.89%)	⬆️
aries_vcx/src/handlers/out_of_band/sender.rs	39.74% <0.00%> (+39.74%)	⬆️

... and 349 files with indirect coverage changes

[Patrik-Stas]

@gmulhearn as a creator of original issue, do you agree this being in a single file? I think your vision to have 2 files, 1 for alice, one for faber.
I am fine with that too. Just started in one file to make this simple for me, but actually might be a lot more readable for someone if we split the roles.

[mirgee]

After integration tests are in better shape after:

• Declutter issuance&presentation protocols #945
• Remove msg-sending IO for issuer and holder #946
• and subsequent removal of IO, and therefore coupling with vcxagency-node mediator
• and some refactoring of "Alice" "Faber" devsetup agents

I believe that integration tests could be in condition to demonstrate the usage (and surface outstanding API issues) equally good as this demo.
But for the time being, this demo is a lot simpler than the existing integration tests.

Will we remove the demo after the work mentioned is completed, i.e. integration tests are in such a condition that they fulfill the current purpose of the demo, i.e. to demonstrate aries-vcx usage, as discussed in speaking?

[Patrik-Stas]

After integration tests are in better shape after:

• Declutter issuance&presentation protocols #945
• IO for issuer and holder as opt-in #946
• and subsequent removal of IO, and therefore coupling with vcxagency-node mediator
• and some refactoring of "Alice" "Faber" devsetup agents

I believe that integration tests could be in condition to demonstrate the usage (and surface outstanding API issues) equally good as this demo.
But for the time being, this demo is a lot simpler than the existing integration tests.

Will we remove the demo after the work mentioned is completed, i.e. integration tests are in such a condition that they fulfill the current purpose of the demo, i.e. to demonstrate aries-vcx usage, as discussed in speaking?

I'd be in favor, there surely will be a lot of duplication between this and integration tests after refactoring them

[gmulhearn]

nit; log colors are printing a little funky for me, is this expected?

(ran with RUST_LOG="info" cargo run)

[gmulhearn]

nit; not sure if you were looking for it , but you could also call PairwiseInfo::create(wallet) to avoid construction of PairwiseInfo at the end

[Patrik-Stas]

I would like us to actually get rid of the whole PairwiseInfo thing, which why I am not using it, rather creating it as necessity at the point where it's needed.
The "did" portion of PairwiseInfo is generally useless

[gmulhearn]

This is really awesome, nice work. preemptive approval from me; just some comments to consider. (assuming others are happy too ocourse)

[gmulhearn]

send_message_2 bugs me a bit too.. maybe something like send_message_as_inviter send_message_as_invitee? not sure

[bobozaur]

Mostly left remarks.

I think the message sending could be improved, but that would require quite a bit of a refactor. The duplicate methods should be easy to remove though, which is why I'd request some changes.

[bobozaur]

Wasn't the previous implementation similar to this in terms of communication? Maybe I'm missing something but I don't get why the shared ownership and interior mutability is needed.

Each agent should be in charge of its own connections (senders to other agents and receivers from other agents) and practically own them exclusively. Or is there some reason why this can't be achieved?

[Patrik-Stas]

Archiving this. Instead, a new demo will be created soon on top of aries-vcx-agent crate