Patch: Potential fix for panic on call to block w/ group resource sharing.

[gmalouf]

No description provided.

[jannotti]

Looks good. I'd kind of like to know why the old code is copying stxn.Txn.Account[k-1] instead of just returning it. But it seem it's being overzealous. If sdk.Address is declared as an array (not slice), this has to be safe.

[codecov]

Codecov Report

Attention: 6 lines in your changes are missing coverage. Please review.

Comparison is base (9ef21ad) 68.49% compared to head (50a40d7) 68.42%.

Files	Patch %	Lines
api/converter_utils.go	68.42%	5 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1590      +/-   ##
==========================================
- Coverage   68.49%   68.42%   -0.07%     
==========================================
  Files          37       37              
  Lines        7427     7434       +7     
==========================================
  Hits         5087     5087              
- Misses       1911     1917       +6     
- Partials      429      430       +1     

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

[algorandskiy]

I wonder if this new logic should be conditional based on protocol version, @jannotti ?

[jannotti]

I'd say it's ok. This is a copy of the same code in go-algorand. My reasoning is that the old would error if the index was out of bounds. This code will too, unless the bug that caused a large index also created a SharedAccts array (which would not exist in old versions). So the behavior should always be the same as it was unless we have two coincidental bugs that happen to compensate for each other. In such a case, we return the account that has somehow made its way into SharedAccts