bug: CRITICAL statement in IF statement triggers "Unexpected end of scope at line"

[han190]

Is there an existing issue for this?

• I have searched the existing issues

Description

As described in the title, when CRITICAL statement appears in IF statement, an error "Unexpected end of scope at line" is triggered.

Screenshots

Expected Behaviour

I believe the test code should be standard conforming.

Version of Modern Fortran

v3.4.2023011101

Version of Visual Studio Code

v1.74.3

Platform and Architecture

WSL

Additional Information

No response

[gnikit]

Thanks @han190, this appears to be a bug in fortls, I will transfer the issue there and issue a fix ASAP. Since you are working with coarrays, if you notice anything else that needs fixing do let me know. I have done very little work with coarrays in fortls so I would expect there would be plenty of issues.

[gnikit]

See PR #256 (will merge once green).

I will wait a bit before I release a new version since the current master of fortls contains a bunch of new features that I need to go over again to make sure I don't disrupt people's workflows.

You can test the new version and tell me if you see anything wrong, including this bug by using:

pip install git+https://github.com/fortran-lang/fortls.git

[han190]

Thank you, the issue is resolved so fast! In #256 I saw that you combined block and critical in the same regex pattern. I guess this is fine for a quick fix, but based on the standard a BLOCK construct is

R1107 block-construct    is block-stmt
                         [ block-specification-part ]
                         block
                         end-block-stmt
R1108 block-stmt         is [ block-construct-name : ] BLOCK

and a CRITICAL construct is

R1116 critical-construct is critical-stmt
                            lock
                            end-critical-stmt
R1117 critical-stmt      is [ critical-construct-name : ] CRITICAL [ ( [ sync-stat-list ] ) ] 

So syntactically speaking a CRITICAL construct may include an extra [ ( [ sync-stat-list ] ) ] . Like what we discussed in #172, many of the coarray issues encountered require a better Fortran parser. And, yes I will keep you posted if I spotted new issues.