February 2006 - ntg-pdftex - NTG Mailing Lists

Re: [NTG-pdftex] [pdftex] pdftex, dvipdfmx, encguide
by Hartmut Henkel 28 Feb '06

28 Feb '06

On Tue, 28 Feb 2006, Hans Hagen wrote: > Werner LEMBERG wrote: > > > > I thought I'd pass on this recent note from Werner about pdftex > > > > generating a much larger file than dvipdfmx. [...] > > > > > > > thanks a lot. Very interesting! This was done with 1.21a, but even > > > the new 1.40.0-beta is not better. > > > > This is the very reason why pdftex is a bad choice for documents > > written in a CJK script. > > > indeed, although i remember that thanh looked into it (or was going to > look into it) > > it should be put high on the agenda it is there, now :-) The /Encoding in the embedded Type1 font file should be independent from an external encoding. Currently the external encoding is duplicated also in the font, unnecessarily, as this info is ignored anyway, due to the full external /Encoding. So we might have similar fonts in the PDF that differ only by their unused encodings. I believe we should keep the original /Encoding inside the font intact (unused anyway, only for information), using the value pairs coming with the font file. We can reduce the size of the /Encoding vector in the font so that only the entries of actually used glyphs are there. E. g. if an external /Encoding says 13 --> /foo and the internal original encoding says 55 --> /foo then if 13 --> /foo is typeset by pdftex, we need to lookup /foo in some hash or tree for the font file so we get the original 55 which goes into the subsetted font file stream. I'll try this. Then there is some magic needed to trigger glyph activation in one PFB file only... Regards, Hartmut

1 0

[ pdftex-Patches-438 ] \pdfsavepos in DVI mode
by noreply＠sarovar.org 28 Feb '06

28 Feb '06

Patches item #438, was opened at 2005-10-16 21:52 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=438&group_id=106 Category: Positioning Group: v1.30.0 Status: Open Resolution: Accepted Priority: 5 Submitted By: Heiko Oberdiek (oberdiek) Assigned to: Hartmut Henkel (hhenkel) Summary: \pdfsavepos in DVI mode Initial Comment: \pdfsavepos in DVI mode Hello, using \pdfsavepos in a package of mine (tabularht) I was wondering, why \pdfsavepos is not allowed to work in DVI mode. \pdfsavepos in PDF mode is based on cur_h and cur_v, both coming from the DVI frontend. The only difference I found is that the PDF format and pdfTeX's PDF mode know about page dimensions. Thus the directions of the axes are the same for the PDF format and the values of \pdfsavepos (\pdflastxpos, \pdflastypos). Origin: lower left corner horizontal axis: values increasing from left to right vertical axis: values increasing from bottom to top unit: sp The DVI case doesn't know about page dimensions, thus I suggest using cur_h=0 and cur_v=0 as origin. It is the left upper corner of the box that is being shipped out. Origin: left upper corner of the box horizontal axis: values increasing from left to right vertical axis: values increasing from bottom to top unit: sp Thus the absolute values differ in both modes and I don't see that this can be avoided. However, if the application uses a reference point then the relative values are exactly the same. The attachted patch enables and implements \pdfsavepos for the DVI mode. It also contains the patch for bug 437 (Two \immediate cancel themselves). Yours sincerely Heiko <oberdiek(a)uni-freiburg.de> ---------------------------------------------------------------------- Comment By: Hans Hagen (hagen) Date: 2006-02-28 00:24 Message: Logged In: YES user_id=927 positioning has always been available in dvi mode as well by means of specials and postprocessing the dvi file (first by a script by taco, later by the dvipos program that comes with dvipdfmx); btw, when you mean 'the current drivers' you refer to latex, and you cannot assume that those are used outside latex, so that is no real reference -) in a similar fashion the dvipos functionality is no real reference either since it is mainly meant for appending info to the context second-pass-info file; anyhow, since pdftex is now the standard engine, there is no need to support the dvipos specials (apart from backward compatibility in context) because we had a "pos:papersize w h" special, dvipos could provide the same reference frame as when using the pdftex primitives; therefore, using the pdfpagewidth and pdfpageheight specification as in the suggested latest patch) is the best way to go (at least for context) (Since we now have pdfpagewidth/height in dvi mode, we get less and less \pdfprimitives that complain in dvi mode) Hans ---------------------------------------------------------------------- Comment By: Heiko Oberdiek (oberdiek) Date: 2005-10-17 11:39 Message: Logged In: YES user_id=1276 Hello, I have now extended the patch: \pdfpagewidth/height are now used for the page dimensions. Also the funny DVI origin offset of 1in is taken into account. Thus the values in DVI and PDF mode are the same. The exception is \pdf{h,v}origin. If the value is not 1in, then the values differ in DVI and PDF mode. However, also the DVI and PDF output will differ in this case. Thus the different values are then ok, I think. If someone wants to have the same values, he just sets \pdfpage{width,height} and let \pdf{h,v}origin untouched. Hartmut Henkel wrote: > to give page dimension info for pdftex also in > DVI mode, one could implement the standard > \special{papersize=x,y} also in pdftex (parallel > to \pdfpagewidth/height), so that it would be > scanned, and also go into the DVI file. Would > this help? I think, not really: * Scanning of \specials are done in PDF mode already, but here it is not really necessary, the current drivers know about \pdfpage{width,height}. * In DVI mode, the scanning of \special's contents must then be added to set \pdfpage{width,height}. But who needs this actually? * Relative positioning with \pdfsavepos works with or without the correct page size information. * Absolute positioning is more difficult in use (If you are somewhere in a box, how do you want to apply absolute positions?) * If someone insist on absolute positioning, he will probably not rely on an accidentally issued \special{papersize=...}. He will probably make sure, that the papersize information is set and then he can also set \pdfpage{width,height}. Yours sincerely Heiko <oberdiek(a)uni-freiburg.de> PS: The name of the patch file is now "v2-pdftex.ch.diff". And I have removed the patch of fixed bug 437. ---------------------------------------------------------------------- Comment By: Hartmut Henkel (hhenkel) Date: 2005-10-16 22:16 Message: Logged In: YES user_id=929 to give page dimension info for pdftex also in DVI mode, one could implement the standard \special{papersize=x,y} also in pdftex (parallel to \pdfpagewidth/height), so that it would be scanned, and also go into the DVI file. Would this help? Regards, Hartmut ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=438&group_id=106

1 0

[ pdftex-Patches-466 ] \primitive and \ifprimitive
by noreply＠sarovar.org 27 Feb '06

27 Feb '06

Patches item #466, was opened at 2006-01-07 17:22 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=466&group_id=106 Category: Primitives >Group: v1.30.0 Status: Open Resolution: Accepted Priority: 4 Submitted By: Taco Hoekwater (taco) Assigned to: Martin Schröder (oneiros) Summary: \primitive and \ifprimitive Initial Comment: This patch implements two new primitives: \primitive \par This will run the 'primitive' command \par, even if it has been re- or undefined in the meantime. \primitive is not expandable, because it may be asked to execute unexpandable primitives as well as expandable ones. If you supply a nonexistant primitive, it will be ignored (no error is raised). The second primitive is a new type of \if: \ifprimitive \par ... \else .. \fi returns true if the current meaning of \par is the 'primitive' command \par. The patch is against alpha-20051226, i hope that is not too old (?). I expect to do at least one minor update during the beta cycle, after the pdftex code is flattened. It was quite a struggle to make this change file work ok, and as a result, there are a few ugly web hacks. ---------------------------------------------------------------------- Comment By: Bernd Raichle (bernd) Date: 2006-02-14 08:17 Message: Logged In: YES user_id=3344 IMHO it would be better to rephrase the above to (and change the implementation accordingly :-(): <quote mode="changed"> \primitive is expandable, because it may asked to execute expandable primitives as well as unexpandable ones. \primitive<cs> will expand to one token, the original TeX primitive. This token will then be processed further as usual. </quote> If \primitive is left unexpandable, a lot of programming techniques seems to be unavailable, e.g., all redefined expandable primitives in expandable-only context. Or any assignment to the get the original primitive, like \edef\orignumber{\expandafter\noexpand\primitive\number} \expandafter\let\expandafter\orignumber\primitive\number Sidenote: The new \primitive seems to implement a subset of a full-blown namespace mechanism. If (pdf)TeX would have namespaces, one simply has to access the appropriate token within the "primitive" namespace (e.g. \primitive::number to access the original \number primitive, assuming the notation \<namespace>::<csname>). How about choosing an appropriate namespace mechanism for (pdf)TeX and implementing it instead? ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2006-02-13 18:56 Message: Logged In: YES user_id=1608 We need to think about this. I see your point, but making \primitive expandable forces the introduction of un-redefinable tokens, and that has its own problematic side-effects (like slowing down each and every definition). I welcome third opinions on whether the benefits outweigh that downside. At one point I tried making the expandability depend on the referred primitive, but that gave me headaches at the implementation stage (a pity, because that would have been nice). Yet another solution is to have an expandable command as well as an unexpandable command (two primitives instead of just the single \primitive). On the side note: I will happily write an implementation ... but only if someone comes up with a specification that works under all conditions instead of only under specific programming styles :-) ---------------------------------------------------------------------- Comment By: Bernd Raichle (bernd) Date: 2006-02-13 17:40 Message: Logged In: YES user_id=3344 IMHO it would be better to rephrase the above to (and change the implementation accordingly :-(): <quote mode="changed"> \primitive is expandable, because it may asked to execute expandable primitives as well as unexpandable ones. \primitive<cs> will expand to one token, the original TeX primitive. This token will then be processed further as usual. </quote> If \primitive is left unexpandable, a lot of programming techniques seems to be unavailable, e.g., all redefined expandable primitives in expandable-only context. Or any assignment to the get the original primitive, like \edef\orignumber{\expandafter\noexpand\primitive\number} \expandafter\let\expandafter\orignumber\primitive\number Sidenote: The new \primitive seems to implement a subset of a full-blown namespace mechanism. If (pdf)TeX would have namespaces, one simply has to access the appropriate token within the "primitive" namespace (e.g. \primitive::number to access the original \number primitive, assuming the notation \<namespace>::<csname>). How about choosing an appropriate namespace mechanism for (pdf)TeX and implementing it instead? ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=466&group_id=106

1 0

[ pdftex-Patches-438 ] \pdfsavepos in DVI mode
by noreply＠sarovar.org 27 Feb '06

27 Feb '06

Patches item #438, was opened at 2005-10-16 23:52 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=438&group_id=106 Category: Positioning >Group: v1.30.0 Status: Open Resolution: Accepted Priority: 5 Submitted By: Heiko Oberdiek (oberdiek) Assigned to: Hartmut Henkel (hhenkel) Summary: \pdfsavepos in DVI mode Initial Comment: \pdfsavepos in DVI mode Hello, using \pdfsavepos in a package of mine (tabularht) I was wondering, why \pdfsavepos is not allowed to work in DVI mode. \pdfsavepos in PDF mode is based on cur_h and cur_v, both coming from the DVI frontend. The only difference I found is that the PDF format and pdfTeX's PDF mode know about page dimensions. Thus the directions of the axes are the same for the PDF format and the values of \pdfsavepos (\pdflastxpos, \pdflastypos). Origin: lower left corner horizontal axis: values increasing from left to right vertical axis: values increasing from bottom to top unit: sp The DVI case doesn't know about page dimensions, thus I suggest using cur_h=0 and cur_v=0 as origin. It is the left upper corner of the box that is being shipped out. Origin: left upper corner of the box horizontal axis: values increasing from left to right vertical axis: values increasing from bottom to top unit: sp Thus the absolute values differ in both modes and I don't see that this can be avoided. However, if the application uses a reference point then the relative values are exactly the same. The attachted patch enables and implements \pdfsavepos for the DVI mode. It also contains the patch for bug 437 (Two \immediate cancel themselves). Yours sincerely Heiko <oberdiek(a)uni-freiburg.de> ---------------------------------------------------------------------- Comment By: Heiko Oberdiek (oberdiek) Date: 2005-10-17 13:39 Message: Logged In: YES user_id=1276 Hello, I have now extended the patch: \pdfpagewidth/height are now used for the page dimensions. Also the funny DVI origin offset of 1in is taken into account. Thus the values in DVI and PDF mode are the same. The exception is \pdf{h,v}origin. If the value is not 1in, then the values differ in DVI and PDF mode. However, also the DVI and PDF output will differ in this case. Thus the different values are then ok, I think. If someone wants to have the same values, he just sets \pdfpage{width,height} and let \pdf{h,v}origin untouched. Hartmut Henkel wrote: > to give page dimension info for pdftex also in > DVI mode, one could implement the standard > \special{papersize=x,y} also in pdftex (parallel > to \pdfpagewidth/height), so that it would be > scanned, and also go into the DVI file. Would > this help? I think, not really: * Scanning of \specials are done in PDF mode already, but here it is not really necessary, the current drivers know about \pdfpage{width,height}. * In DVI mode, the scanning of \special's contents must then be added to set \pdfpage{width,height}. But who needs this actually? * Relative positioning with \pdfsavepos works with or without the correct page size information. * Absolute positioning is more difficult in use (If you are somewhere in a box, how do you want to apply absolute positions?) * If someone insist on absolute positioning, he will probably not rely on an accidentally issued \special{papersize=...}. He will probably make sure, that the papersize information is set and then he can also set \pdfpage{width,height}. Yours sincerely Heiko <oberdiek(a)uni-freiburg.de> PS: The name of the patch file is now "v2-pdftex.ch.diff". And I have removed the patch of fixed bug 437. ---------------------------------------------------------------------- Comment By: Hartmut Henkel (hhenkel) Date: 2005-10-17 00:16 Message: Logged In: YES user_id=929 to give page dimension info for pdftex also in DVI mode, one could implement the standard \special{papersize=x,y} also in pdftex (parallel to \pdfpagewidth/height), so that it would be scanned, and also go into the DVI file. Would this help? Regards, Hartmut ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=438&group_id=106

1 0

[ pdftex-Patches-434 ] Remove queued map item handling
by noreply＠sarovar.org 27 Feb '06

27 Feb '06

Patches item #434, was opened at 2005-10-13 01:08 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=434&group_id=106 Category: Fonts >Group: v1.30.0 Status: Open Resolution: Accepted Priority: 5 Submitted By: Hartmut Henkel (hhenkel) Assigned to: Taco Hoekwater (taco) Summary: Remove queued map item handling Initial Comment: Remove queued map item handling from mapfile.c ============================================== Formerly all items from \pdfmapfile{+-=mapfile} or \pdfmapline{+-=mapline} were queued up in a list until the next font was selected; then the whole list was processed, e. g. all queued mapfiles or maplines read in. When one gave \pdfmapfile{mapfile} or \pdfmapline{mapline} without +-=, the pending list was cleared. The only practical use for this was to prevent reading in the default mapfile. But this can be done simpler, without queuing. With this patch all \pdfmapfile{+-=mapfile} or \pdfmapline{+-=mapline} requests are processed completely and immediately without queuing, which means that if e. g. a \pdfmapfile{+-=mapfile} primitive is scanned, the mapfile will be read in right away. This happens independently from selecting a font. When \pdfmapfile{+-=mapfile} or \pdfmapline{+-=mapline} is scanned, and there is a default mapfile, it is read first and then the mapfile. Reading of the default mapfile can be disabled by \pdfmapfile{} or \pdfmapline{} or \pdfmapfile{mapfile} or \pdfmapline{mapline} without +-= as long as the default mapfile hasn't already been read in. Latest when the first font is requested, the default mapfile will be read, if this hasn't already been disabled by the mentioned primitives. Apart from possibly reading the default mapfile, there is no other action when a font is requested, since the \pdfmapfile{+-=mapfile} or \pdfmapline{+-=mapline} primitives have already acted right away when they were called. The effects of this patch are the following: * Initial mapfile reading still can be blocked in the usual way. * The difficult to describe queuing mechanism has gone, code gets simpler. * Mapfile reading happens directly by the primitives. * Therefore it's possible to use \pdfretval with the primitives, to check success of the mapfile/line reading (not yet implemented). * As the return value is connected to the \pdfmapfile or \pdfmapline call, one can easier implement other map setting primitives and give them return values. This was not possible before, as the queuing was inbetween. * For any standard use, the patch should be compatible to the former queuing method. The mapfile.c is a patched version of the mapfile.c from Thanh's subfont patch pdftex-20050924.patch.gz for pdftex-1.30.3. Hope it doesn't sound too messy :-) For discussion. Regards, Hartmut ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=434&group_id=106

1 0

[ pdftex-Patches-432 ] Rules in a PDF qQ group
by noreply＠sarovar.org 27 Feb '06

27 Feb '06

Patches item #432, was opened at 2005-10-06 21:01 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=432&group_id=106 Category: Positioning >Group: v1.30.0 Status: Open Resolution: Accepted Priority: 5 Submitted By: Hartmut Henkel (hhenkel) Assigned to: Hans Hagen (hagen) Summary: Rules in a PDF qQ group Initial Comment: Put rules with their positioning on the page inside a qQ group. Should make \pdfliteral page {...} behave consistently after text or rules or images, as then any positioning starts from the PDF origin; no delta-positioning anymore. It's already being tested in pdfluatex. Just to have it at sarovar... Testing welcome, particularily with text and graphics transforms by graphicx package, to spot backward incompatibilities, if any... Regards, Hartmut ---------------------------------------------------------------------- Comment By: Hartmut Henkel (hhenkel) Date: 2005-10-09 15:07 Message: Logged In: YES user_id=929 Update pdf_set_origin_temp: Now has (h, v) parameters. Regards, Hartmut ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=432&group_id=106

1 0

[ pdftex-Patches-427 ] Character positioning patch
by noreply＠sarovar.org 27 Feb '06

27 Feb '06

Patches item #427, was opened at 2005-09-11 12:39 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=427&group_id=106 Category: Positioning >Group: v1.30.0 Status: Open Resolution: Accepted Priority: 5 Submitted By: Hartmut Henkel (hhenkel) Assigned to: Hans Hagen (hagen) Summary: Character positioning patch Initial Comment: Experimental patch increases resolution of /Widths array by 1 fractional digit, makes corrections of type ")-1(" and ")1(" in the TJ array happen much less often, particularily for CM fonts. Gives e. g. 1 % PDF file size shrink for pdfetex web sources compiled into PDF. Rough patch only, not for production. Testing welcome. Regards, Hartmut ---------------------------------------------------------------------- Comment By: Hartmut Henkel (hhenkel) Date: 2005-10-09 15:06 Message: Logged In: YES user_id=929 Patch extended: Now pdf_set_origin has (h, v) parameters, simplifies pdf_begin_text procedure. ---------------------------------------------------------------------- Comment By: Hartmut Henkel (hhenkel) Date: 2005-10-09 00:25 Message: Logged In: YES user_id=929 Removed pdf_first_space_corr variable, as it seems that it was defunct anyway: There was no pdf_end_string procedure call (only place where something is done with pdf_first_space_corr), where afterwards the pdf_h variable (then modified by pdf_first_space_corr) would actually be used. E. g. pdf_moveto uses pdf_last_h instead, and after pdf_end_text there is anyway a pdf_set_origin call. And pdf_begin_string gets simpler. Testing welcome! Regards, Hartmut ---------------------------------------------------------------------- Comment By: Hartmut Henkel (hhenkel) Date: 2005-10-03 01:16 Message: Logged In: YES user_id=929 patch updated: now the TJ array building in procedure pdf_begin_string is simplified/tidied up. Testing welcome! Regards, Hartmut ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=427&group_id=106

1 0

[ pdftex-Patches-416 ] new primitive pdflastlink
by noreply＠sarovar.org 27 Feb '06

27 Feb '06

Patches item #416, was opened at 2005-08-19 09:42 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=416&group_id=106 Category: Primitives >Group: v1.30.0 Status: Open Resolution: Accepted Priority: 5 Submitted By: Ralf Utermann (ralfu) Assigned to: Hartmut Henkel (hhenkel) Summary: new primitive pdflastlink Initial Comment: This patch implements a new primitive \pdflastlink (analogous to \pdflastannot) which returns the object number of the last object created with \pdfstartlink. ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=416&group_id=106

1 0

[ pdftex-Patches-415 ] various extensions and fixes
by noreply＠sarovar.org 27 Feb '06

27 Feb '06

Patches item #415, was opened at 2005-08-19 09:09 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=415&group_id=106 Category: Primitives >Group: v1.30.0 Status: Open Resolution: Accepted Priority: 5 Submitted By: The Thanh Han (hanthethanh) Assigned to: Martin Schröder (oneiros) Summary: various extensions and fixes Initial Comment: This is a patch for various experimental extensions: - \pdfinsertht: return the height of an insert (requested by Hans); this has similar syntax to \pdftexrevision. - do not fail when \pdfobj is used with an invalid object number (requested by Hans) - snapping: support for grid typesetting -- see snap.txt for details - kern before space: allow changing interword glues in a flexible way and inserting a kern before certain characters -- see kern-before-interword-glue.txt for details The notes in snap.txt and kern-before-interword-glue.txt are from email correspondence with people interested in; hence these notes are somewhat informal, inaccurate or out-of-date. How to apply: ~~~~~~~~~~~~~ cd /path/to/pdftex-1.30.0-srcdir/src/texk patch -p1 < /path/to/pdftex-20050818.patch cd web2c make pdfetex ---------------------------------------------------------------------- Comment By: Martin Schröder (oneiros) Date: 2005-09-04 18:22 Message: Logged In: YES user_id=421 The patch for invalid object numbers is included in 1.30.1++ ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=415&group_id=106

1 0

[ pdftex-Patches-386 ] grow obj_tab and dest_names arrays dynamically
by noreply＠sarovar.org 27 Feb '06

27 Feb '06

Patches item #386, was opened at 2005-07-10 20:29 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=386&group_id=106 Category: Memory allocation >Group: v1.30.0 Status: Open Resolution: Accepted Priority: 5 Submitted By: Hartmut Henkel (hhenkel) Assigned to: Hartmut Henkel (hhenkel) Summary: grow obj_tab and dest_names arrays dynamically Initial Comment: experimental patch to grow the obj_tab and dest_names arrays dynamically; arrays initially have 1000 entries. This required also update of file avlstuff.c. Similar to patch-366, which did this already for the pdf_mem array. Then all (?) pdftex-specific memory configuration parameters, obj_tab_size, dest_names_size, and pdf_mem_size in texmf.cnf would be ignored. Regards, Hartmut ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=495&aid=386&group_id=106

1 0