Uwe Stöhr wrote: > Am 01.06.2013 13:52, schrieb Stephan Witt: > >> But the documentation in filetools.h clearly states for quote_python: >> /* >> Quote for python. Use this if you want to store a filename in a python >> script. Example: os << "infile = " << quoteName(filename) << '\\n'; >> */ >> >> Here we have a command line argument for a python call to be parsed by a >> shell or a shell similar QProcess call. We do not store the argument in a >> python script. >> >> So it should be quote_shell. Or do I miss something?
I would think the same, but I do not know the quoting works with the QProcess based child process execution. Before QProcess was used, it was simple: quote_python was only used for python scripts, like it is still documented, and qutoe_shell for all file names which were given to the OS shell. Now, there is a home made command line parser which translates some artificial syntax in something that can be used by QProcess, and the question is what this parser expects. > I also understand it that way but there must be a reason that this was > once used. I am not an expert in this field and therefore would no make > any action. git blame is your friend in such cases, or trac: http://www.lyx.org/trac/browser/lyxgit/src/support/Package.cpp?annotate=blame . This tells you that Enrico introduced the quote_python argument in http://www.lyx.org/trac/changeset/c73eb3497aa3e34b2536cf44071053d30d5c5166/lyxgit . Since Enrico is an expert in these things I would use a little bit of cargo cult and keep the quote_python flag. Or better, ask him about the reasons and document them. Georg
