Patches item #368, was opened at 2005-06-18 00:29 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=368&group_id=106 Category: Primitives Group: v1.30.0

Status: Closed Resolution: None Priority: 5 Submitted By: Hartmut Henkel (hhenkel) Assigned to: Nobody (None) Summary: make \pdfstrcmp expandable (experimental patch)

Initial Comment: just moved \pdfstrcmp over to procedure conv_toks, so that the result -1/0/1 is inserted into the input token list. Does this make sense? ----------------------------------------------------------------------

Comment By: Hartmut Henkel (hhenkel) Date: 2005-06-20 22:09

Message: Logged In: YES user_id=929 Patch 368 can be closed as an improved version is now included in patch 371 by Heiko. Many thanks to him and Taco for debugging! Regards, Hartmut ---------------------------------------------------------------------- Comment By: Heiko Oberdiek (oberdiek) Date: 2005-06-20 17:02 Message: Logged In: YES user_id=1276 Hello, Taco I agree with your analysis, last weekend I found the same variables that need saving, too. Also I have a patch that also make \pdfescape* expandable. Thus \pdflastescaped* are not needed anymore. And I could remove the dependency on a limited buffer by copying the destination strings directly in the string pool above the source string. Third I added \pdfescapehex and \pdfunescapehex that adds encoding/decoding for hexadecimal strings: * Strings can also be given as hex strings in angle brackets. * Safe method to store and read string data to auxiliary files without worrying about catcodes, unmachted curly braces, ... * Help in handling UTF-8/Unicode or other hex coding stuff. I hope, I can attach the patch here in this thread (upload part is missing?. Otherwise I will make an new patch. Patch is applied by patch ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2005-06-20 16:33 Message: Logged In: YES user_id=1608 It makes sense, but it is not 100% safe as-is. In an expanding context, lots of stuff can go one at the same time. Esp., scan_toks touches some global variables, so compare_strings should save the globals def_ref, scanner_status and warning _index to prevent stack corruption later on. For instance, this crashes: % process by initex \input plain \edef\alpha{\pdfstrcmp{alpha}{alph}} \message{\meaning\alpha} \end % I am attaching a change file that modifies compare_strings() , cf. scan_general_text() in eTeX. ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=368&group_id=106