Closed tomwhite closed 5 months ago
If I remember correctly, in the AWS Lambda backend this behavior was intended so that the runtime name passed to the list_runtime()
can act like a filter in the way *runtime_name*
.
I agree the runtime to be deleted should be only that runtime that exactly matches the name passed to the delete command, but in my opinion the fix should be placed in the delete_runtime()
function of the cli, in here:
Thanks @JosepSampe.
If I remember correctly, in the AWS Lambda backend this behavior was intended so that the runtime name passed to the
list_runtime()
can act like a filter in the way*runtime_name*
.
Other backends don't seem to have wildcard matching in list_runtimes()
. For example:
So it would be more consistent to make AWS Lambda do exact matches for list_runtimes()
too don't you think?
(Also, the lithops runtime list
CLI always shows all runtimes - there's no filtering or wildcard matching going on there either.)
Yes, I agree. Someone was using list_runtimes()
programmatically in the AWS Lambda backend in the past (not trough the cli), but for now we can proceed like this
When I ran
it didn't just delete the
cubed-runtime
image, but anything withcubed-runtime
in the name, such ascubed-runtime-dev
,cubed-runtime-main
, etc. This is clearly a bug.Previously I reported a bug where it wasn't possible to delete any AWS Lambda runtimes (#1290), where I suggested that runtime name matching should use equality:
But the fix in #1294, did not use equality, but instead used inclusion (
in
):