[ pdftex-Feature Requests-385 ] \pdfnoligatures should leave kerning intact
Feature Requests item #385, was opened at 2005-07-10 17:38 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=496&aid=385&group_id=106 Category: None Group: None
Status: Closed Resolution: Postponed Priority: 5 Submitted By: Robert (schlcht) Assigned to: Taco Hoekwater (taco) Summary: \pdfnoligatures should leave kerning intact
Initial Comment: % The new \ptexnoligatures primitive will break the % horizontal positioning of \mathaccents if the base % character is slanted. % There is no problem with \accent, nor with \mathaccent % on upright characters: $\dot{A} \dot{\rm{A}} \dot{\Delta} \dot{\mit\Delta}$ {\tenit \accent'137 A} \ptexnoligatures\tenit % OK \ptexnoligatures\textfont0 % = cmr10, \rm % OK \ptexnoligatures\textfont1 % = cmmi10, \mit % breaks shift $\dot{A} \dot{\rm{A}} \dot{\Delta} \dot{\mit\Delta}$ {\tenit \accent'137 A} \bye, Robert ----------------------------------------------------------------------
Comment By: Martin Schröder (oneiros) Date: 2006-07-25 00:14
Message: Logged In: YES user_id=421 I think further effort on this is wasted, as \ptexnoligatures will be reimplemented with OpenType fonts anyway, i.e. in the near future. So we should just document the current behaviour and carry on. ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2006-02-20 17:50 Message: Logged In: YES user_id=1608 Maybe we should just document the weird behaviour then and keep the thing marked experimental? I do not have lots of time to spend on this hack right now, nor do I want to, since everything I do now will have to be thrown out again for version 1.50. The change file that is attached to this feature item can then be ignored, except for that one line that re-enables \tagcode. ---------------------------------------------------------------------- Comment By: Bernd Raichle (bernd) Date: 2006-02-20 17:32 Message: Logged In: YES user_id=3344 Looking at Taco's change file, it should be possible to break his "non-vmode fix" by inserting \vbox{\pdfnoligatures\font} or \vadjust{\pdfligatures\font} in hmode. A final fix is to reimplement this \pdfnoligatures primitive: Either do not only change the |tag| field in the char_info for all font glyphs, but remove the ligs from the lig/kern program. Or save the information that no ligature should be constructed within the hlist (using a special node to start and end the nolig part similar to the setlanguage whatsit nodes or by adding the nolig flag to the character nodes). ---------------------------------------------------------------------- Comment By: Robert (schlcht) Date: 2006-02-20 16:40 Message: Logged In: YES user_id=2217 Ah, I didn't realize it had such confusing effects. However, I still believe that disallowing it in non-vmode is too restrictive. ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2006-02-20 09:56 Message: Logged In: YES user_id=1608 But it is not global for the current paragraph, it is only applied to words that are looked at for hyphenation. Try this test, and have a close look at the first and third word of the paragraph. There are even stranger possibilities when \language is taken into account. \font\tst=cmr10 \hsize=2in \noindent \tst efficiency efficiency \hbox{efficiency} \pdfnoligatures\tst efficiency efficiency efficiency \bye ---------------------------------------------------------------------- Comment By: Robert (schlcht) Date: 2006-02-19 19:19 Message: Logged In: YES user_id=2217 In my opinion, this latest patch is too restrictive. It should be sufficient to document \pdfnoligatures as one of those commands whose scope is global to the current paragraph (and the rest of the document), just like setting, say, \hyphenpenalty or \hsize in the middle of a paragraph affects the complete paragraph. The current behaviour is also consistent with that of the hz-related stuff: \rpcode and friends have a retroactive effect, too. (Besides, the patch would break microtype.sty, which sets up the fonts as they are being loaded, which would happen in horizontal mode, mostly... ;-) PS. If it was for me, this feature request could be closed; I totally accept the "known limitation". ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2006-02-15 10:58 Message: Logged In: YES user_id=1608 The attached patch disallows \pdfnoligatures and \tagcode in non-vmodes, and restores the assignment functionality of \tagcode (that was an unrelated bug in 1.40-beta) ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2006-02-15 10:17 Message: Logged In: YES user_id=1608 We can disallow the primitive in non-vertical mode. Makes sense. ---------------------------------------------------------------------- Comment By: Bernd Raichle (bernd) Date: 2006-02-15 10:10 Message: Logged In: YES user_id=3344 Another limitation which should be documented: If \ptexnoligatures<font> is applied within a paragraph, the ligs and kerns created and inserted before this command are removed depending if a hyphenation pass is necessary to justify the paragraph. \ptexnoligatures<font> has the _global_ side effect to suppress all ligs and kerns of the font which are created _afterwards in time_. TeX breaks the lines of a paragraph at the end of the paragraph. If hyphenation has to be applied, ligs are remapped to single chars, (implicit) kerns are removed ... and reconstituted afterwards using the font information valid at this (later) time. If the font info is changed in between, which is true if \ptexnoligatures<font> is applied, already inserted ligs and kerns are removed. ---------------------------------------------------------------------- Comment By: Martin Schröder (oneiros) Date: 2005-07-12 22:23 Message: Logged In: YES user_id=421 Changing this into a feature request. ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2005-07-10 18:50 Message: Logged In: YES user_id=1608 This is because it (\ptexnoligatures) also disables the kerning information for the font. You'll have to accept that as a "known limitation" for now. ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=496&aid=385&group_id=106
participants (1)
-
noreply@sarovar.org