Created attachment 26756 [details] testcase After the upgrade from 10.04.0 to 10.05.0 and now 10.05.1, files generated by ps2pdf are larger. It seems to be due to duplicate fonts from the original PDF file that are kept duplicate. I could produce a simple testcase with \documentclass[11pt]{article} \usepackage[T1]{fontenc} \usepackage{lmodern} \begin{document} \thispagestyle{empty} \begin{itemize} \item T $=$ \end{itemize} \end{document} compiled using "pdflatex duplicate-fonts.tex". I've attached the generated PDF file, for which pdffonts gives: name type encoding emb sub uni object ID ------------------------------------ ----------------- ---------------- --- --- --- --------- ABKCER+LMRoman10-Regular Type 1 Custom yes yes yes 4 0 ABKCER+LMRoman10-Regular Type 1 Custom yes yes yes 5 0 ABKCER+LMRoman10-Regular Type 1 Custom yes yes yes 6 0 When I use ps2pdf with Ghostscript 10.04.0 (Debian package), pdffonts gives: name type encoding emb sub uni object ID ------------------------------------ ----------------- ---------------- --- --- --- --------- SFKWXB+LMRoman10-Regular Type 1C Custom yes yes yes 7 0 But when I use ps2pdf with Ghostscript 10.05.1 (Debian package), pdffonts gives: name type encoding emb sub uni object ID ------------------------------------ ----------------- ---------------- --- --- --- --------- YZFINB+LMRoman10-Regular Type 1C Custom yes yes yes 7 0 ULUEGO+LMRoman10-Regular Type 1C WinAnsi yes yes yes 9 0 FBBLED+LMRoman10-Regular Type 1C WinAnsi yes yes yes 11 0 and this PDF file is larger than the previous one: With Ghostscript 10.04.0: 4056 bytes With Ghostscript 10.05.1: 6252 bytes I did not try this testcase with Ghostscript 10.05.0, but it already produced larger PDF files: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1102891 (I hadn't identified the cause at that time), so I suppose that it is this bug.
This was a bug in 10.04.0 which was fixed with a series of commits. Too many to track down individually, but starting with 3276619a7eb5b2269a158f115a163847be31458e It is entirely possible that the bug in 10.04.0 was not causing issues with your use case, but it was still a bug, caused problems for other users, and has been fixed. Try testing against (for example) the 9.55.0 release, which was the last one using the old PostScript-based PDF interpreter. You will find it behaves the same way as 10.05.1. Note that the fonts are not *duplicates* because they have different encodings. The original input file also has 3 fonts defined in it, they simply all use the same FontDescriptor.