Intel® Fortran Compiler
Build applications that can scale for the future with optimized code designed for Intel® Xeon® and compatible processors.
29263 Discussions

suggestion for error message #6633 The type of the actual argument differs from the type of the dumm

gauthier_108
Novice
807 Views

I often come across this error message from the type checker:

 

>  error #6633: The type of the actual argument differs from the type of the dummy argument. [symbolname]
call some_routine(symbolname)
---------------------^

 

I wanted to suggest the message to be improved:

 

it would:

* state the two types that are mismatched

* provide the locations of type declaration (or first place the implicit symbol is declared)

 

This would make the troubleshooting much more effective, and also the compiler more user friendly.

 

What do you think?

 

Is there a sense that the compiler frontend (or whole), would be opensourced? as this is typically the kind of issue the community can take over.

 

F# community has seen lot of engagement and contributions to the compiler in similar fashion:

https://github.com/dotnet/fsharp/issues/1103

0 Kudos
1 Solution
Ron_Green
Moderator
773 Views

The improvements you suggest are good.  I will open a Feature Request for these. 

 

On the topic of open sourcing:  no, that will not happen.  We have discussed and debated this extensively and it would make no sense.  The complexity and primarily the age of the front end make it intractable and unattractive to master. We would never get a community to build and rally around it. 

View solution in original post

4 Replies
Ron_Green
Moderator
774 Views

The improvements you suggest are good.  I will open a Feature Request for these. 

 

On the topic of open sourcing:  no, that will not happen.  We have discussed and debated this extensively and it would make no sense.  The complexity and primarily the age of the front end make it intractable and unattractive to master. We would never get a community to build and rally around it. 

gauthier_108
Novice
729 Views

Ron, this is great!

 

Sorry, I guess the question about the compiler implementation has been asked countless times!

 

Is there some sense though, that community oriented activity could be leveraged in contributing to the documentation?

 

I find the reference to be very formal and a bit too terse.

 

Something that is in between "user has specific issue and posts on forum" and "formal specification" with finest technical writing requirements could be useful for people the use the compiler and ecosystem.

 

I'm thinking a bit like how MSDN was centralized and eventually became to some extent distributed / crowd sourced.

 

For example, C# compiler messages are described, for some, to great extent:

 

https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/compiler-messages/feature-version-errors

 

This is something that community may be able to contribute to.

0 Kudos
Devorah_H_Intel
Moderator
670 Views

We have had many requests to improve error messages submitted via the Service Center in addition to the forum. 

Fortran Developer Guide and Reference also contain info on different types of error messages. 

0 Kudos
Ron_Green
Moderator
657 Views

The feature request for improvement to error #6633 has ID CMPLRLLVM-65611

 

 

 

0 Kudos
Reply