> > Am 17.09.2015 um 03:08 schrieb David Wright <lily...@lionunicorn.co.uk>: > > > > I wrote earlier: > >>>> After all, in your script you have: > >>>> pdfcrop --margins 1 '$file_base_name.pdf' '$file_base_name.pdf' > >>>> where the output filename is the same as the input filename. > > > > Well I'm still worried about the line I quoted above, which I think is > > inviting trouble (unless you clearly understand the mechanism that is > > being used). Bear in mind, in my script I used an intermediate file: > > $ pdfcrop --margins 1 foo.pdf bar.pdf
Quoting poto...@posteo.de (poto...@posteo.de): > Apparently I don't clearly understand the mechanism. > Why can't the program write the cropped pdf to the same file? When you're lucky, the program will read a small input file and, if you're really lucky, close it before opening the output file, doing the processing and then writing the output. In general, programs aren't designed to do that because it limits the quantity of input to what will fit in memory. pdfcrop is designed to handle arbitrarily large input files and, indeed, it will handle input from a pipe (which means it has no idea how much input there will eventually be). Also consider what happens if you press ^C halfway through the process: you've destroyed the input file. That's dealing with only a tiny part of the file-handling implications, and doesn't even consider file-locking on some OSes which will prevent your opening a file for writing when it's already open for reading. But now look at it from a naive point of view. You assume that less data has been written than has been read at *all* times during the process (as well by the end). Can you guarantee that? These filesizes might surprise you (they did me): Input file, the moderately-sized Lua Reference Manual, obviously output from the TeX family 474818 refman-5.0.pdf Output from $ pdfcrop --margins 1 refman-5.0.pdf via-file 4709449 via-file Output from $ pdfcrop --margins 1 - via-pipe < refman-5.0.pdf 4709378 via-pipe So with your script, you're likely to come a cropper one day. (Sorry, had to get that in!) If you're going to write a script, always use the most conservative assumptions, not short-cuts. And before you try to debug a script with a possible race-condition, you need to remove any obvious potential sources for the problem. Cheers, David. _______________________________________________ lilypond-user mailing list lilypond-user@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-user
