Open kba opened 2 years ago
Is it possible to provide an example for ambiguous behavior?
I think this epics lacks a step when deprecating --mets-basename
: We should update this parameter's docs to make clear how its behaviour can be achieved by using the other two params. Or is this already implicit?
Is it possible to provide an example for ambiguous behavior?
For example, if you do ocrd workspacd -d /path/to/workspace1 -m /path/to/workspace2/mets.xml -M not-mets.xml
, where should we look for the METS?
/path/to/workspace2/mets.xml
(because of -m
)/path/to/workspace1/not-mets.xml
(because of -d
and -M
)
Current situation
The three options
-M/--mets
,-m/--mets-basename
and-w/--working-dir
(processors) /-d/--directory
(ocrd workspace
) are inter-connected and they can clash to cause ambiguous behavior when combined.For example, if you do
ocrd workspacd -d /path/to/workspace1 -m /path/to/workspace2/mets.xml -M not-mets.xml
, where should we look for the METS?/path/to/workspace2/mets.xml
(because of-m
)/path/to/workspace1/not-mets.xml
(because of-d
and-M
)How it should be
--mets-basename
should be deprecated inocrd workspace
, it can be simulated with a combination of--mets
and--directory
.--working-dir
should be properly taken into account for processor calls.The resolving logic of processors and
ocrd workspace
should ideally use the same code for consistencySteps
-mets-basename
could be dropped