Support old stata -b mode where stdout was written to a log file. #8
Conversation
It may be that users are include the current log file stuff in their outputs.
entrypoint.sh
Outdated
|
|
||
| # batch mode writes output for script.do to script.log, so we preserve that | ||
| # behaviour | ||
| log=${script%.do}.log |
There was a problem hiding this comment.
I think Stata uses just the script filename and not the full path to determine the log file, so the log file always get written in the cwd regardless of where the script is:
https://www.statalist.org/forums/forum/general-stata-discussion/general/1354640-can-i-control-the-location-of-log-file-in-batch-mode
There was a problem hiding this comment.
Of course it does.
Fixed
There was a problem hiding this comment.
If you want to know how Stata behaves you can just ask yourself, "How could Stata make my life more difficult in this situation?"
It occurs to me, btw, that both this and the wrapped.$script thing might have been caught by the tests if the test scripts ran from a subdir of the cwd.
There was a problem hiding this comment.
It occurred to me too, which is why I added in the last commit :)
Unless you meant something different?
There was a problem hiding this comment.
🤦 Totally missed that! Sorry! Though great minds etc
Also, move the test scripts into a subdirect to ensure we test using directories in the path properly
It may be that users are include the current log file stuff in their
outputs.