.do files of the form do/a.do were not found #3
Conversation
It should produce exactly the same result whether a .do file is under a do/ dir or not (e.g. ./do/y.do or ./y.do), but it wasn't the case, for example: $ cat ./do/default.do # empty file $ cat ./do/x.do redo y $ redo y # builds ./y redo y $ redo x # builds ./x and ./do/y redo x redo do/y The last command should build ./x and ./y, not ./do/y.
|
This pull request has a bug, due to misusing rstrip. In three places in state.py, it calls ".rstrip('/do')" as though rstrip was a right-aligned replace, when it really has more in common with 'tr -d'. The reason the parameter is named 'chars' is because it isn't a string at all - it's a set of chars which rstrip is permitted to treat as 'trailing garbage'. This has the amusing effect of causing redo to fail in building itself, because it chops off the 'do' and then can't find 're'. |
|
One way to resolve that: abs_pwd = abs_pwd[0:-3] if abs_pwd[-3:] == '/do' else abs_pwd |
|
I pushed some changes in the dodir feature where |
First commit:
With a 'do/default.do' file, or 'do/a.do' file, the command 'redo a.do' did not work.
Note: in the test 107-dodir, the line 'mkdir -p x/y' seems unnecessary.
Commits 2 & 3: tiny cleanups