summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAge
* Don't bother overriding the metrics of PostScript fonts. If the metricsBen Harris2004-09-25
| | | | | | | we've got are wrong, things are going to look horrible either way, so we may as well apply minimal effort. [originally from svn r4580]
* Fix first two bugs in compound text support: escape sequences wereSimon Tatham2004-09-25
| | | | | | | mis-ordered, and initial charset state failed to specify 8859-1 in GR. [originally from svn r4579] [this svn revision also touched charset,filter,timber]
* The COMPOUND_TEXT encoding used by some X applications to transferSimon Tatham2004-09-25
| | | | | | | | | | | | | internationalised text in selections is a subset of ISO 2022 containing no base character sets which libcharset doesn't already support. As such, it isn't too hard to add direct compound text support into libcharset, so here it is. With any luck I should eventually be able to integrate this into Unix PuTTY, to deal with the fact that the useful Xutf8 functions we currently use are specific to XFree86. [originally from svn r4578] [this svn revision also touched charset,filter,timber]
* If you're going to change the metrics of a font, it's necessary to leave spaceBen Harris2004-09-24
| | | | | | | in the new font dictionary for the Metrics entry. This is mentioned in the first edition of the Red Book, but not the second. [originally from svn r4577]
* Replace the empirical GhostScript-derived font metrics with ones derivedBen Harris2004-09-24
| | | | | | | | | | from Adobe's official AFM files. These are expressed in integer multiples of 1/1000 pt (well, 1/1000 of the design size of the font, actually, but that's always a whole number of points), so use that as our internal unit rather than 1/4096 pt. This doesn't seem to make a significant difference to Halibut's output. [originally from svn r4576]
* Page-break penalties were calculated by taking the amount of spare spaceBen Harris2004-09-24
| | | | | | | | | on the page (in internal units) and squaring it. This was fine except that fixed penalties weren't scaled by the size of an internal unit, so the page- breaking changed when the units were changed. Rather than scaling all the fixed penalties, scale the space into 1/4096 point units before squaring it. [originally from svn r4575]
* Stupid bug: "t" was leaving the y-coordinate on the stack, which confusedBen Harris2004-09-21
| | | | | | | "restore". Quite how GhostScript managed not to give an error on this I don't know. [originally from svn r4565]
* Replace the hardcoded instances of "4096" and "4096.0" in the paper backendsBen Harris2004-09-21
| | | | | | | with a macro. halibut.ps and halibut.pdf are identical (modulo dates) over this change. [originally from svn r4564]
* Bracket each page in a save/restore pair. This is suggested by the Red Book,Ben Harris2004-09-21
| | | | | | | enforces page independence, avoids leaking VM on level 1 interpreters, and speeds things up to boot. [originally from svn r4561]
* Make the PostScript prologue look less like it was written by a C programmer.Ben Harris2004-09-20
| | | | | | | | In particular, use the operand stack to hold "x" and "y", and use a dictionary lookup to switch based on the type of "x". This also seems to give a slight speed increase. [originally from svn r4553]
* Two small tweaks to the prologue:Ben Harris2004-09-20
| | | | | | | | | * Use "/arraytype" etc rather than "[] type" etc. Should be faster and not fill local VM with empty arrays. * Use "bind" on the procedure, since it's conventional and will probably help speed too. [originally from svn r4552]
* Add lots of DSC comments, and change the DSC version to 3.0, which has beenBen Harris2004-09-20
| | | | | | | | | | | | | current since 1990. The only obvious change from this is that gv now displays the document title. There's a slight bug here at the moment in that the backend emits %%DocumentNeededResource and %%IncludeResource for each subfont of a PostScript font, even if that PostScript font has been requested already. This is wasteful but, as far as I can see, not actually disallowed, and is easier than de-duping the font list. [originally from svn r4551]
* Use "$(MAKE)" for sub-makes, so as to handle systems whose default "make"Ben Harris2004-09-19
| | | | | | isn't GNU Make. [originally from svn r4548]
* Forgot to set a file position on characters being returned fromSimon Tatham2004-08-20
| | | | | | macro expansions. [originally from svn r4489]
* Remove a \#{FIXME} comment in \cfg{paper-chapter-top-space}; the checkinJacob Nevins2004-08-09
| | | | | | | of bk_paper.c 1.29 [r4330] appears to have fixed this. [originally from svn r4432] [r4330 == 8aac8532019ac6cee701d48ceb35473280219a0e]
* `make release' tweak to prevent tarball containing trailing NULs (fix fromJacob Nevins2004-08-09
| | | | | | putty/mkunxarc.sh) - apparently this upsets some older gzips (1.2.4). [originally from svn r4431]
* Explicitly bless in the documentation, and fix in the man backend,Simon Tatham2004-08-06
| | | | | | | the practice of using \dt and \dd in anything other than the obvious interleaving. [originally from svn r4417]
* Make sure (paragraph)->private_data is defined in all cases, on generalJacob Nevins2004-08-06
| | | | | | | principles (it may be unnecessary). Perhaps we should do this for (word)->private_data too. [originally from svn r4412]
* Fix crash in Info backend: if a \[Kk] to a \B or \n was followed in aJacob Nevins2004-08-06
| | | | | | paragraph by another \[Kk], Bad Things would happen. [originally from svn r4411]
* dup_word_list() does the right thing if passed NULL (on general principles)Jacob Nevins2004-08-06
| | | | [originally from svn r4410]
* Fix for an uninitialised structure member spotted by valgrind (probably didn'tJacob Nevins2004-08-04
| | | | | | have any real effect). [originally from svn r4401]
* Explicit bounds checking on the bmp[] array.Simon Tatham2004-08-04
| | | | [originally from svn r4400]
* Add an explicit cast to prevent a compiler warning.Simon Tatham2004-08-04
| | | | [originally from svn r4399]
* Colin spotted a sizeof/lenof confusion which we think has beenSimon Tatham2004-08-04
| | | | | | causing segfaults on strange architectures. [originally from svn r4398]
* More careful context hash calculation which doesn't depend on sizeSimon Tatham2004-07-30
| | | | | | of `unsigned long'. Should help Debian build problems (bug#259504). [originally from svn r4377]
* Division by zero (which criminally failed to give rise to any kindSimon Tatham2004-07-17
| | | | | | | of signal!) was causing massive misplacement of the single line of text on a one-line page in PS and PDF. [originally from svn r4330]
* Cut-and-paste error was preventing PS and PDF filename configSimon Tatham2004-07-17
| | | | | | options from being processed in some circumstances. [originally from svn r4329]
* Tweak for \cq (probably unnecessary)Jacob Nevins2004-07-16
| | | | [originally from svn r4328]
* Right; I'm finally sick of typing \q{\cw{foo}}, so I've invented aSimon Tatham2004-07-16
| | | | | | shorthand command \cq{foo}. [originally from svn r4327]
* Failed to initialise state.curr_topic early enough, causingSimon Tatham2004-07-13
| | | | | | | uninitialised-memory references and a segfault. Quite why this never happened when I tested it at home, I have no idea! [originally from svn r4326]
* Apparently (according to a user) WinZip complains of `trailingSimon Tatham2004-07-05
| | | | | | | | | garbage' when handling a GNU tar archive. Using the -o flag when building the tar file creates an archive which the user said WinZip had no trouble with, and GNU tar seems just as happy with that, so `make release' now uses -o. [originally from svn r4323]
* Avoid generating an index section if there actually is no index.Simon Tatham2004-07-02
| | | | [originally from svn r4321]
* Sort out error handling everywhere a charset name is converted intoSimon Tatham2004-06-27
| | | | | | an integer charset ID. [originally from svn r4317]
* Free up all the data we allocated during the HTML backend.Simon Tatham2004-06-27
| | | | [originally from svn r4316]
* Support non-breaking spaces.Simon Tatham2004-06-27
| | | | [originally from svn r4315]
* Config directives for all the extra configurable things in the newSimon Tatham2004-06-27
| | | | | | HTML backend. [originally from svn r4314]
* Fix up the restrict_charset and output_charset configuration for theSimon Tatham2004-06-27
| | | | | | | | | | | | | | HTML backend. After an argument with James in which it transpired that we each thought it was absolutely obvious which one \cfg{html-charset} ought to map to, and each of us was thinking of a different one, I've decided that this implies it _isn't_ obvious, and hence there is no single \cfg{html-charset} directive at all. Instead, we have \cfg{html-restrict-charset} and \cfg{html-output- charset}, and any naive user who `just wants to change character set' is just going to have to RTFM and figure out which one they mean. [originally from svn r4313]
* nit: s/case-sensitive/case-insensitive/ in comment about nasty indexingJacob Nevins2004-06-21
| | | | | | hack. [originally from svn r4311]
* Various HTML backend cleanups.Simon Tatham2004-06-20
| | | | [originally from svn r4308]
* Introduce a configurable option to select the HTML flavour. AlsoSimon Tatham2004-06-20
| | | | | | | fiddle with various small aspects of the output so that it actually validates in all supported flavours. [originally from svn r4307]
* Improve the naming of fragment IDs for numbered list elements andSimon Tatham2004-06-20
| | | | | | bibliography entries. [originally from svn r4305]
* Minimal documentation of all the new paper backend \cfg{} directives.Jacob Nevins2004-06-20
| | | | | | | | | Contains many FIXMEs, both for incomplete documentation and incomplete code. Couple of other minor tweaks elsewhere. [originally from svn r4304]
* We seem to have ended up with duplicate documentation of --man. Gone.Jacob Nevins2004-06-19
| | | | [originally from svn r4303]
* Ensure that every \cfg{} directive documented so far has appropriate \IMJacob Nevins2004-06-19
| | | | | | directives. [originally from svn r4302]
* Fairly wholesale reorganisation of the \cfg{} documentation for the text,Jacob Nevins2004-06-19
| | | | | | | | WinHelp, man, and info backends, and organisation into sections (now that we have rather more directives documented). I've not changed the actual text much if at all. [originally from svn r4301]
* In the text backend, multiple characters can now be used for underlining.Jacob Nevins2004-06-19
| | | | [originally from svn r4300]
* The Unicode OVERLINE in the text backend documentation was probably supposedJacob Nevins2004-06-19
| | | | | | to be quoted. [originally from svn r4299]
* Note that \cfg{quotes} and friends apply to the output in its entirety andJacob Nevins2004-06-19
| | | | | | can't be changed partway through. Also, add a cross-reference from \q{}. [originally from svn r4298]
* Note that no index is generated by the text backend.Jacob Nevins2004-06-19
| | | | [originally from svn r4297]
* Minor correction to --ps and --pdf descriptions.Jacob Nevins2004-06-19
| | | | [originally from svn r4296]
generated by cgit v1.1 at 2026-04-27 21:23:58 +0000