Fix empty name variable on RHEL9

[ajm-ska]

Description

This PR fixes #1250

It is similar to issue #1199 so I used the same idea to modify line 811 src/Region/CrtfImportExport.cc from

if (name.front() == '"' && name.back() == '"') {

to

if (!name.empty() && name.front() == '"' && name.back() == '"') {

With this change, the CASA_REGION_IMPORT_EXPORT test passes on RHEL9 again.

I notice there are similar lines in Ds9ImportExport.cc and RegionImportExport.cc, but I have not touched them as there are no other issues with the ICD tests at present. I'm not sure if there could be potential problems in future when the use of the newer GCC (>11.2.1) becomes more common?

Checklist

• [x] changelog updated / no changelog update needed
• [x] e2e test passing / added corresponding fix
• [x] no protobuf update needed
• [ ] added reviewers and assignee
• [ ] added ZenHub estimate, milestone, and release

[ajm-ska]

@kswang1029 Do you think this warrants a changelog entry?

[kswang1029]

@kswang1029 Do you think this warrants a changelog entry?

No changlog is needed I think.

[confluence]

I notice there are similar lines in Ds9ImportExport.cc and RegionImportExport.cc, but I have not touched them as there are no other issues with the ICD tests at present.

I think that use of front / back without a check that the string is not empty is a logic error which is bound to cause problems later, and we should fix this everywhere while we're fixing it here.

[ajm-ska]

I think that use of front / back without a check that the string is not empty is a logic error which is bound to cause problems later, and we should fix this everywhere while we're fixing it here.

OK. On closer inspection, the front() and back() call in Ds9ImportExport.cc can not be an empty string, so that file does not need modification: if (name.front() == '{' && name.back() == '}') {

However, RegionImportExport.cc does have the possibility to call empty strings, so I have applied the same ! .empty() logic fix to the appropriate instances in that file.

As far as I can see, there are no other instances in the source code where there could be calls to empty strings, so this should be the end of the problem.

I have tested the region import/export ICD tests on macOS and RHEL9 on this branch and they continue to pass.

[confluence]

On closer inspection, the front() and back() call in Ds9ImportExport.cc can not be an empty string

I'm not sure about this -- the usage seems directly analogous to the CRTF import/export file. The string is looked up in a map just like it is here, but that doesn't mean the value can't be empty.

[github-actions[bot]]

Package	Line Rate	Health
src.Cache	68%	➖
src.DataStream	52%	➖
src.FileList	68%	➖
src.Frame	51%	➖
src.HttpServer	43%	➖
src.ImageData	28%	❌
src.ImageFitter	89%	✔
src.ImageGenerators	52%	➖
src.ImageStats	74%	➖
src.Logger	44%	➖
src.Main	54%	➖
src.Region	19%	❌
src.Session	30%	❌
src.Table	52%	➖
src.ThreadingManager	87%	✔
src.Timer	85%	✔
src.Util	50%	➖
Summary	39% (6818 / 17596)	➖

[ajm-ska]

I'm not sure about this -- the usage seems directly analogous to the CRTF import/export file. The string is looked up in a map just like it is here, but that doesn't mean the value can't be empty.

OK. I have added the same ! .empty() check to the line.

I have run the import/export ICD tests (they include DS9-related tests) on macOS and RHEL9 again and they continue to pass, so everything should be still fine.