[pdftex] Status of \pdfadjustspacing

[Hartmut Henkel]

On Wed, 19 Nov 2003, John Culleton wrote:

> When I attempted to use the \pdftajustspacing feature it seemed not to
> work. When I set up my body font with a statement like:
> \font\rm= bchr8r at 9pt stretch 30 shrink 20 step 10
>
> ....the extra parameters were simply printed as text.
>
> Does anyone have any additional information on this feature?

Seems there are a few problems: First, kpathsea complains about TFM font
names with numeric (expanded) extensions "Illegal fontname...contains
'+'" (e. g. cmr12+50, which would mean that the TFM is expanded by
50/1000 in width). This seems to require a patch in
texk/kpathsea/tex-make.c, function kpse_make_tex. Did this. Then I could
not find any automatic means of generating the expanded mf-sources to
create the tfm's. Did this by hand, patched copies of cmr12.mf:
cmr12-100.mf, cmr12-50.mf, cmr12+50.mf..., using the lines from Thanh's
dissertation in the TUGBoat, Vol. 21 No. 4, pg. 372 (I use extreme
expansion to see what happens). Now lots of tfm's are created, e. g.
cmr12+50.tfm, cmr12+100.tfm. But then pdftex-1.11b crashes with a
segfault. Why? Accidentally I had made a patch in these areas, and,
dont't ask me, but the patched version happily succeeds and gives a PDF
file with expanded fonts, but pixel fonts only. Looks funny, like these
modern magazines where character stretch sometimes is almost 2:1. But
this is only due to my crazy settings. The test file:

\pdfcompresslevel=0\hsize=100mm\nopagenumbers
\newcount\N\def\resetefcode#1{ \N=0\loop\efcode#1 \N=1000
\advance\N by1 \ifnum\N<256\repeat}\pdfadjustspacing=2
\font\cmr cmr12 at12pt\resetefcode\cmr
\pdffontexpand\cmr 300 300 100 1000
\cmr\input tufte\bye

So, loose ends seem about to be the kpathsea patch, automatisms to
create patched mf's (they must exist, but where?), the automatic
modification and renaming of non-pk fonts by changing their /FontMatrix
to fake the Expand by an Extend, the segfault in pdftex.

Greetings, Hartmut