Example 14 has traditional been excluded from the non-interactive tests since it outputs 2 files, and you can only specify one file on the command line using the -o option. With a judicious redirection of stdin it is possible to work round this. I've implemented this for all languages so example 14 is now part of the standard ctest suite. As a consequence I've found and corrected several differences between implementations. This has brought 2 issues to light
1) In order to get this to work for devices which produce single pages (e.g. png) I've had to modify the C version to call plgfam on stream 1 and then plsfam on stream 2 with the read parameters. This works fine, except that the command line option -fflen which sets the number of digits to use for the family page number has no API equivalent that I can find. It is not part of plgfam / plsfam which it probably should be. This means that the file names for the second stream are things like x14ac1.ps rather than x14c01.ps. As a result of this I've not yet propagated the plgfam / plsfam changes to other languages. 2) A wider question is if and how command line options should be passed on to subsequent streams. It might be possible to call plparseopts with the second stream to re-parse the command options, but this would results in both streams having the same output name. This is going to depend on the application on a case-by-case basis I suspect, but we should at least supply the user with the relevant plgxxx / plsxxx functions to check and set anything set on the command line. 3) The other major issue that this all brought to light is that the xfig driver is broken for use with multiple streams. It uses several static variables in the driver for storing stream-specific settings which will obviously mess things up. Most seriously it uses a malloc'ed buffer. Multiple calls to plend1 for the multiple streams will result in a crash. This results in example 14 crashing with the xfig test. This can be relatively easily fixed, but I probably won't get time today, so if someone else gets there first, then feel free to take a look. Andrew ------------------------------------------------------------------------------ _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel
