ppmraw -r72 01_001.pdf : In the big text "571-L" the vertical stem in 'L' is wider than in 'L', but GS renders the first one narrower than the second one. Detected by Julia as a regression from December 09, 2003. Reproduced by Igor with today's CVS HEAD. The problem happens due to two stem boundaries are grid-fitted independently. Use the stem width instead StdVW.
Created attachment 806 [details] 01_001-.pdf A reduced test
A correction : "the vertical stem in 'L' is wider than in '1'".
The bug appears invalid due to an inaccurate analyzis. Actually the design width of the stem in '1' is 71966, and the design width of thew stem in 'L' is 75766 in the gxhintn.c units. Ghostscript renders '1' wider than 'L', so the result is correct.
Oops, the previous comment appears wrong. The width of '1' is 71936, lesser than the width of 'L' 75776. GS renders '1' wider, the result is incorrect.
Created attachment 881 [details] ccc-.pdf One more useful test made from comparefiles/ccc.pdf . See the character 'i' at 72 dpi. It has 2 vertical stems with a common left side, which alings with a left shift. If we do so, entire glyph looks shifted to the left.
Created attachment 884 [details] patch.txt The patch waits for approval.
Patch http://www.ghostscript.com/pipermail/gs-cvs/2004-September/004793.html
Now Julia reports a degradation of the height of the character 's' since the June 09 revision with ppmraw -r72 045-01.ps
Patch http://www.ghostscript.com/pipermail/gs-cvs/2004-September/004881.html