ippeveprinter and pagesize tolerance

[jubork]

ippeveprinter (2.4.2) does not allow for any pagesize tolerance.
I'm not sure whether IPP standard allows for any.
The relevant code is (ippeveprinter.c:8779)

      for (i = 0; i < count ; i ++)
      {
        size  = ippGetCollection(supported, i);
        x_dim = ippFindAttribute(size, "x-dimension", IPP_TAG_ZERO);
        y_dim = ippFindAttribute(size, "y-dimension", IPP_TAG_ZERO);

        if (ippContainsInteger(x_dim, x_value) && ippContainsInteger(y_dim, y_value))
          break;
      }

The problem comes from the fact that IPP specifies page size in 0.01mm units,
while PPDs usually in integer postscript points (1/72 inch). The conversion between the two
may introduce small offset, and a print job submitted to ippeveprinter may
be rejected during verify phase.

The quick&dirty (very little testing done) patch below implements
more robust conversion algorithm, which seems to work for me
for a couple of printers/PPDs.

cups-2.4.2-pspoints.patch.txt

However, I think better approach would be to allow for some page size tolerance,
i.e. via command line option. The default could be zero for compatibility.

On a side note, the patch also improves handling of IPP_TAG_ENUM options.
Without it, conversion of enums to integer gives often zero (again very little
testing done).

[michaelrsweet]

@jubork There is no hard requirement for media size tolerance, although the current guidance says the following:

A requested media size dimension matches a supported media dimension if it is within an implementation-defined tolerance. For example, PostScript specifies a tolerance of 5 points (5/72 of an inch = 1.7 mm) of a supported dimension, i.e., within 176 units of the value of the dimension.

The PWG media functions in libcups use a tolerance of 1mm.

[michaelrsweet]

Oh, and reviewing the patch we really couldn't use it - the encode.c changes would break existing clients, and I'm not sure how all the rounding/truncation you are doing will affect things (particularly for metric sizes).

The correct fix should allow for a tolerance range when checking...

[michaelrsweet]

[master e2fb1f4] Allow media-size values to match within +/-1mm (Issue #487)