dialects: Raise a parsing error in accfg.setup when result type is invalid

[math-fehr]

Stacked PRs:

• core: Make IRDLOperation results typed #3991
• ->dialects: Raise a parsing error in accfg.setup when result type is invalid #3990
• core: Make OpResult and BlockArgument generic like SSAValue #3989
• core: Add support for typed SSA values in isa #3988
• core: Make SSAValue generic #3987
• core: Move SSAValue definitions around (NFC) #3986

---

dialects: Raise a parsing error in accfg.setup when result type is invalid

As there are a redundancy in the assembly format between the first
parsed string and the result type, this should be a parsing error.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 90.19%. Comparing base (04eeaec) to head (1be5202).

Additional details and impacted files

@@                  Coverage Diff                  @@
##           math-fehr/stack/4    #3990      +/-   ##
=====================================================
+ Coverage              90.16%   90.19%   +0.02%     
=====================================================
  Files                    458      458              
  Lines                  58315    58317       +2     
  Branches                5693     5694       +1     
=====================================================
+ Hits                   52581    52600      +19     
+ Misses                  4341     4322      -19     
- Partials                1393     1395       +2     

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

[alexarice]

Seems reasonable though I'm not really sure why the operation parses/prints a type at this point if it's fully determined by the other inputs.

[JosseVanDelm]

Thanks @math-fehr !

[JosseVanDelm]

Seems reasonable though I'm not really sure why the operation parses/prints a type at this point if it's fully determined by the other inputs.

@alexarice, are you suggesting that you would expect to print/parse it as:

%state = accfg.setup "acc1" to ("A" = %one : i32)

[alexarice]

Yes

[alexarice]

To expand a bit, I wouldn't necessarily expect the custom format to be a certain way (as in I don't feel it's wrong at the moment), but in this case it doesn't feel like the type adds much information to the user, and doesn't seem to be needed by the system.

[JosseVanDelm]

@alexarice yeah no worries, it's a nice suggestion :) thanks!

[compor]

Where is %zero used? Am I missing something?