jmlapre/sst-core

Remove straggler deprecated function calls

#23

· jmlapre opened

on Oct 4, 2024

Os_simple_command() : should we replace this with OSCommand from test_engine_support? Or just tell ppl to create their own subprocess runners if they need to execute something? Or leave as is?

#21

· jmlapre opened

on Sep 24, 2024

Os_wc() : It would be more intuitive to return (line count, word count, char count) instead of the roundabout index shenanigans it is doing right now.

#20

· jmlapre opened

on Sep 24, 2024

Os_awk_print() : Can we get rid of this?

#19

· jmlapre opened

on Sep 24, 2024

testing_stat_output_diff() : Is there a way (or should we even try) to unify this with the testing_compare*() functions?

#17

· jmlapre opened

on Sep 24, 2024

testing_compare_* : There are four of these functions where we could have one that takes output file, reference file, filter list, and type of diff to do (exact, sorted, subset)

#16

· jmlapre opened

on Sep 24, 2024

testing_merge_mpi_files() – For stdout/stderr redirects where the filename & directory structure depends entirely on the OpenMPI version

#15

· jmlapre opened

on Sep 24, 2024

combine_per_rank_files() - For files like stat files that are named basename_X.ext where X is the rank ID)

#14

· jmlapre opened

on Sep 24, 2024

Combining MPI files is complicated but useful to keep this in the framework since different OpenMPI versions do it differently. We have two badly named functions – probably rename so their names are intuitive.

#13

· jmlapre opened

on Sep 24, 2024

log*(): Mostly used/useful. No distinction exists afaik between VERBOSE_NORMAL and VERBOSE_LOUD in the framework despite there being separate verbosity levels. Should we eliminate unused levels or leave them?

#12

· jmlapre opened

on Sep 24, 2024

Move away from lightweight wrappers and implementing minor function variants in place of adding function arguments

#5

· jmlapre opened

on Sep 24, 2024

Get_test_runtime_sec() is used internally. It’s not strictly a “get”, it modifies the run time variable so should ONLY be called by run_sst and never externally. Change to internal?

#4

· jmlapre opened

on Sep 24, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Remove straggler deprecated function calls

Os_simple_command() : should we replace this with OSCommand from test_engine_support? Or just tell ppl to create their own subprocess runners if they need to execute something? Or leave as is?

Os_wc() : It would be more intuitive to return (line count, word count, char count) instead of the roundabout index shenanigans it is doing right now.

Os_awk_print() : Can we get rid of this?

testing_stat_output_diff() : Is there a way (or should we even try) to unify this with the testing_compare*() functions?

testing_compare_* : There are four of these functions where we could have one that takes output file, reference file, filter list, and type of diff to do (exact, sorted, subset)

testing_merge_mpi_files() – For stdout/stderr redirects where the filename & directory structure depends entirely on the OpenMPI version

combine_per_rank_files() - For files like stat files that are named basename_X.ext where X is the rank ID)

Combining MPI files is complicated but useful to keep this in the framework since different OpenMPI versions do it differently. We have two badly named functions – probably rename so their names are intuitive.

log*(): Mostly used/useful. No distinction exists afaik between VERBOSE_NORMAL and VERBOSE_LOUD in the framework despite there being separate verbosity levels. Should we eliminate unused levels or leave them?

Move away from lightweight wrappers and implementing minor function variants in place of adding function arguments

Get_test_runtime_sec() is used internally. It’s not strictly a “get”, it modifies the run time variable so should ONLY be called by run_sst and never externally. Change to internal?

Issues

Search results