aboutsummaryrefslogtreecommitdiffstatshomepage
Commit message (Collapse)AuthorAgeFilesLines
* remove unused enum merr;Ingo Schwarze2010-05-241-57/+1
| | | | ok joerg@
* If a .Bl -tag lacks a -width, skip all non-It macros for the implicitJoerg Sonnenberger2010-05-241-3/+6
| | | | | | calculation. E.g. .Sm can occur as direct child of .Bl. OK schwarze@
* fix an obvious oversight introduced in rev. 1.5:Ingo Schwarze2010-05-241-2/+2
| | | | | MANDOCERR_BODYLOST is not intended to be fatal; required to unbreak the OpenBSD build
* Fix 1.125: Add missing return in termp_nm_pre.Joerg Sonnenberger2010-05-221-1/+2
|
* Test .SH with only empty paragraph. Don't test .IP without body.Joerg Sonnenberger2010-05-221-0/+3
|
* lines containing blank characters FIXED in OpenBSD,Ingo Schwarze2010-05-181-1/+12
| | | | and add one new .Bl -tag issue reported by Joerg
* Increase performance by stashing the list type in struct mdoc_node.Kristaps Dzonsons2010-05-176-197/+175
| | | | | | | This will eventually be used so that mdoc_macro can known whether to dump list line arguments into the body (`Bl -column' overflowing). Remove a2list() and arg_listtype() because of this.
* Added tech@ mailing list.Kristaps Dzonsons2010-05-171-2/+9
|
* Enable the unified error/warning enumeration in mandoc.h that'sKristaps Dzonsons2010-05-1731-563/+565
| | | | | | | | | | | stringified in main.c. Allow `An' to handle an argument and child (with a warning). Allow `Rv' and `Ex' to work without a prior `Nm' as groff does (with a warning). Allow inconsistent column syntax to only raise a warning.
* Multiple empty .PPs don't add a vspace eitherJoerg Sonnenberger2010-05-171-0/+3
|
* Entry for .AT and .UCJoerg Sonnenberger2010-05-171-4/+11
|
* Add support for .AT. Properly implement .UC. Add regress tests.Joerg Sonnenberger2010-05-1720-9/+163
|
* Tests for sane and insane handling of lines before and after .SH/.SSJoerg Sonnenberger2010-05-172-0/+23
|
* groff uses three vspaces before the footer too.Joerg Sonnenberger2010-05-171-1/+3
|
* Documented ie/el.Kristaps Dzonsons2010-05-174-11/+41
| | | | | Installing roff.7 and added it to index.sgml. Small fix for `D1' and `Bd' in -Thtml.
* Use a default prefix of /usr/local. Honour DESTDIR for install targets.Joerg Sonnenberger2010-05-171-19/+20
|
* The function term_flushln() had effectively forked in OpenBSD.Ingo Schwarze2010-05-171-45/+42
| | | | | | | | | | | | | | | | This is the bsd.lv part of the main step to bring it back in sync. At the same time, this prevents trailing whitespace in the output: We delay writing blanks until we are sure printable characters follow. This is achieved by * remembering the end of the word instead of its length * and not using vbl any longer for the control of line breaking * such that vbl can sum up all kinds of white space * before writing a word, printing all the blanks collected in vbl * within the word, adding NBSP chars to vbl, then continuing with the word * after the word, adding blanks to vbl, then starting the next word "looks good" kristaps@
* Full support for ie/el. This completes the initial roff support.Kristaps Dzonsons2010-05-1725-91/+331
| | | | | Added test files for ie/el. Using `if 0' as a baseline for "false" roff instructions instead of `if t'.
* Oops: forgot to make roff_cblock() understand the new macros.Kristaps Dzonsons2010-05-171-2/+17
|
* libroff now intelligently throws away `am', `ami', `am1', `de', `dei',Kristaps Dzonsons2010-05-172-19/+86
| | | | and `de1'. This is also documented in roff.7.
* libroff now is aware of which sub-nodes should be interpreted even in aKristaps Dzonsons2010-05-174-83/+179
| | | | | | | | | | conditional-negative context so as to preserve structural integrity. Initial "rules" (deny/allow) are now inherited for future work in evaluating conditionals. Lint-fix noted by Joerg. Added regression tests for zany constructs.
* Add a test case for nesting of .RS/.REJoerg Sonnenberger2010-05-161-0/+16
|
* `ig' support in all its glory. TryKristaps Dzonsons2010-05-1612-34/+294
| | | | | | | | | | | | | | | | | | .ig ig asdf .ig fdsa .. or .ig if asdf .if n \ foo for a laugh. It all works. Lots of regression tests supporting this and documentation for the same.
* Add a basic regression test for .Bl -bullet and the various interactionsJoerg Sonnenberger2010-05-161-0/+64
| | | | with .Bd.
* Regression tests in place for `.if' in libroff.Kristaps Dzonsons2010-05-1612-7/+260
| | | | | Check against some strange `.if' constructs I missed. Added initial roff.7 manual.
* Fully skip first and last line for the purpose of cmp.Joerg Sonnenberger2010-05-161-3/+6
|
* Skip first line when comparing output.Joerg Sonnenberger2010-05-161-1/+3
|
* mandoc should be a variable.Kristaps Dzonsons2010-05-161-2/+2
| | | | Consensus is on using a *.in for regression tests (symmetry-broken by Ingo).
* Remove temp files on success.Joerg Sonnenberger2010-05-161-0/+1
|
* Add a basic regression testing script.Joerg Sonnenberger2010-05-161-0/+39
|
* Re-admit the simple case of `.ig' that works with groff's stranger invocations.Kristaps Dzonsons2010-05-161-124/+72
|
* Support for the general case of `if', which will boil out into otherKristaps Dzonsons2010-05-161-50/+78
| | | | | | | | constructs: single-level and multi-level blocks; nested, varied-level blocks; recalculating buffer (begin) boundaries pushed correctly into the back-ends, and so on. This will allow, once conditions are turned on, for things like `.ie n .TH FOO 1' then `.el .Dd Mdocdate'. Read it and weep.
* Allow roff_parseln() to be re-run.Kristaps Dzonsons2010-05-167-106/+213
| | | | | | | | | Allow roff_parseln() to manipulate the line buffer offset. This is used in situations like `.ie n .TH FOO 1' or `.ie n .ie n', where the line buffer offset is recalculated then the roff parser re-run. Fix mdoc_parseln() and man_parseln() to accept the initial line offset. WARNING: backed-out ALL roff macros whilst accomodating for how roff handles multi-line conditionals (in short, re-running the parser).
* sync to OpenBSD:Ingo Schwarze2010-05-163-5/+7
| | | | | | introduce a #define to get rid of the magic number describing the ASCII character used to represent non-breaking space; ok kristaps
* Fix allowing silly '\'' control character.Kristaps Dzonsons2010-05-163-6/+6
|
* Removed `.' entry from -man hashtable.Kristaps Dzonsons2010-05-151-7/+5
|
* Remove `am', `ami', `de', `dei', and `.' from -man, as they're now in the ↵Kristaps Dzonsons2010-05-158-167/+13
| | | | roff preprocessor.
* Backed-out warning messages (lots).Kristaps Dzonsons2010-05-152-5/+7
|
* The `am', `ami', `de', and holy `dei' are all being properly ignored.Kristaps Dzonsons2010-05-151-36/+24
|
* The `ig' now supports `ig end-macro'.Kristaps Dzonsons2010-05-153-11/+92
| | | | Initial warning/error messages in place (still experimental).
* Pull `ig' out of -man and leave it the roff preparser.Kristaps Dzonsons2010-05-158-30/+11
|
* Magic fix for bogus manuals making -man puke: I don't care if it's magic ↵Kristaps Dzonsons2010-05-151-2/+2
| | | | because roff instructions will be nixed from the underlying systems anyway.
* Segfault fix and letting warnings slip through for now.Kristaps Dzonsons2010-05-152-5/+7
|
* Add mandoc.h, which will eventually replace compilers' inner enum errors ↵Kristaps Dzonsons2010-05-154-26/+421
| | | | | | with unified ones. Add initial roff pre-processor shim, a compiler sitting outside of the other compilers that processes pure roff instructions.
* non-numeric section headers in -man merged to OpenBSDIngo Schwarze2010-05-151-4/+0
|
* Mishandling of : in expr(1)Joerg Sonnenberger2010-05-151-0/+2
|
* Lint-fix for enum return (ok schwarze@).Kristaps Dzonsons2010-05-151-2/+2
|
* For .IP groff requires a single space only after the head, adopt.Joerg Sonnenberger2010-05-151-3/+1
|
* Fixed memory leak introduced in last checkin of man_action TH parsing (only ↵Kristaps Dzonsons2010-05-152-5/+6
| | | | | | | if TH is called again, which never happens anyway). Fixed my email address in man_action.c and msec.c. Removed superfluous <time.h> inclusion in msec.c.
* LIBRARY can also occur in section 9.Kristaps Dzonsons2010-05-159-45/+44
| | | | | | All manual sections (unknown, 3p, 3f, etc.) correctly handled by -mdoc. Useful warning printed if unknown manual section. Checking for manual sections (e.g., LIBRARY) checks only first character, so 3p, 3f, etc. are free.