Fix: zero amount order bug

[lemunozm]

Description

There are two casuistics where an order can contain a zero amount. Such an order can not be closed, and theoretically, it should not exist.

• Added test cases that generate the error. All newly added tests fail without the fix.
• Added fix bypassing creating orders when they have 0 amount.

Q: Should we allow to close zero orders even though they should not be possible now?

[lemunozm]

We do not really need this test if we have the pallet-swaps tests. But I wanted to force a use case from FI PoV. We can finally remove it once we're sure it's fully solved.

[mustermeiszer]

Thanks!! @lemunozm do you think this fix has to live here or we could think about fixing it on the pallet-order-book side?

I am also not sure why having a zero order lead to a missing FI state. Do you know the reason for that or in general explain what the issue was? I mean pallet-order-book does fulfill zero orders, right?

[codecov]

Codecov Report

Attention: Patch coverage is 88.88889% with 2 lines in your changes are missing coverage. Please review.

Project coverage is 48.56%. Comparing base (da63bf1) to head (c21a170).

❗ Current head c21a170 differs from pull request most recent head 00229e4. Consider uploading reports for the commit 00229e4 to get more accurate results

Files	Patch %	Lines
pallets/swaps/src/lib.rs	88.88%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1816      +/-   ##
==========================================
+ Coverage   48.53%   48.56%   +0.03%     
==========================================
  Files         167      167              
  Lines       13314    13320       +6     
==========================================
+ Hits         6462     6469       +7     
+ Misses       6852     6851       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lemunozm]

do you think this fix has to live here or we could think about fixing it on the pallet-order-book side?

I've added a couple of test for cancel zero amount orders and fulfill zero amount orders and they work fine. So from order-book perspective I think everything is fine.

My reason for putting this in pallet-swaps was: if you do not need nothing to swap, you do not need to create an empty order, so pallet swaps should check those cases.

I am also not sure why having a zero order lead to a missing FI state.

Do you mean to this error right? I'm pretty sure it is because when creating a zero-order, from FI PoV, the swap is fully done, and it detects it can clean the associated info. So later, when you fulfill the zero order, you can not find such a state. This should no longer happen because we're blocking the creation of zero-amount orders now.

[mustermeiszer]

Makes sense! Thanks