-common_args problem

[jammer]

::Win10pro64-2004-CHCP 65001
Object: NOT have to resubmit processing tags for every -stay_open submission.
Question: is the -common_args values restricted to only exiftool I/O or Processing arguments?

Question is based on the following test results;
-with "-Description" on -common_args everything works,
  except a "No file specified" error is thrown at the end. Without "-Description"  error is not thrown.
-tested with multiple submissions, and error is thrown after the last -execute, as if it's on the -stay_open
  or false literals.
If I've missed something in the docs let me know.

Thanks in advance
jon

PS D:>cat exiftoolpipe.txt

D:/image.jpg
-execute516
-stay_open
False


PS D:\> exiftool -stay_open true -@ D:/exiftoolpipe.txt -common_args -r -n -json -Description
[{
  "SourceFile": "D:/image.jpg",
  "Description": "this is image.jpg description"
}]
{ready516}
No file specified


PS D:\> exiftool -stay_open true -@ D:/exiftoolpipe.txt -common_args -r -n -json
[{
  "SourceFile": "D:/image.jpg",
  "ExifToolVersion": 12.01,
  "FileName": "image.jpg",
  "Directory": "D:/",
  ...deleted 16 lines
  "BitsPerSample": 8,
  "ColorComponents": 1,
  "ImageSize": "5105 6792",
  "Megapixels": 34.67316
}]
{ready516}

[Phil Harvey]

Hi Jon,

If you are only executing one command, then -execute is not necessary.  After you say -stay_open false, the command continues with the remaining arguments.  In the special case that there are no more arguments, no warning is given, but otherwise a warning is given if the final command has a problem.

If you know that this will be the last file processed, then don't use -execute.

But you raise a good point.  If you don't know beforehand that you will be processing the last file, and you are using -common_args, then I can think of no way to terminate ExifTool without a warning if it is waiting in a -stay_open argfile.  Hmm...  I could maybe introduce a new option like -abort, but it seems like a bit much just to avoid the "No file specified" warning.

What do you think?

- Phil

[jammer]

How would -last, -final or -finalize or something in that nature sound?

[Phil Harvey]

Sounds maybe better.  Or -exit or -end maybe.  But my question was: Do you really think it would be worthwhile to add a new option just to avoid this warning message?

- Phil

[jammer]

Phil:
Is it worth it? Obviously I'm not promoting a breaking change or significant
code change.  If the rest of the community has no concerns, then leave it as is.

I'll do a workaround by creating a fake submission before
the "-stay_open/nfalse/n".  Maybe you could make a note in step 3
of the "-stay_open FLAG" documentation that the final submission
is not to be followed with an -execute command.

My concerns:
1)  throwing an error which has to be handled
2) negates the use of {ready[NUM]}
3) you may not know it's the last submission set.
    IE; use case of GUI where the user requires info from last submission to make the
    determination of finalization.

I do really appreciate your "retirement time" and consideration.
Thanks again
jon

[Phil Harvey]

Hi Jon,

Yes, my biggest concern is the negation of {ready[NUM]} if you want to avoid the warning.

I'll think about this.

- Phil