[BACK]Return to TODO CVS log [TXT][DIR] Up to [cvsweb.bsd.lv] / mandoc

Diff for /mandoc/TODO between version 1.20 and 1.169

version 1.20, 2010/07/02 10:43:21 version 1.169, 2014/04/23 16:08:33
Line 1 
Line 1 
 ************************************************************************  ************************************************************************
   * Official mandoc TODO.
   * $Id$
   ************************************************************************
   
   ************************************************************************
   * crashes
   ************************************************************************
   
   - The abort() in bufcat(), html.c, can be triggered via buffmt_includes()
     by running -Thtml -Oincludes on a file containing a long .In argument.
     Fixing this will probably require reworking the whole bufcat() concept.
   
   ************************************************************************
 * missing features  * missing features
 ************************************************************************  ************************************************************************
   
 - explicit blocks with missing end macro should be implicitely closed  --- missing roff features ----------------------------------------------
   at the end of the enclosing block, e.g. .Bl It (El) Sh  
   reminded by stsp@  in net/pptp pptp.8  Fri, 23 Apr 2010 20:32:39 +0200  
   
 - fix bad block nesting involving multiple identical explicit blocks  - .ad (adjust margins)
   see the OpenBSD mdoc_macro.c 1.47 commit message    .ad l -- adjust left margin only (flush left)
     .ad r -- adjust right margin only (flush right)
     .ad c -- center text on line
     .ad b -- adjust both margins (alias: .ad n)
     .na   -- temporarily disable adjustment without changing the mode
     .ad   -- re-enable adjustment without changing the mode
     Adjustment mode is ignored while in no-fill mode (.nf).
   
 - we need real .ds; for example, terminfo.3 has  - .fc (field control)
   .ds n 5    found by naddy@ in xloadimage(1)
   and then later  
   \fBterminfo\fR(\*n)  - .nr third argument (auto-increment step size, requires \n+)
   reported orally by halex@ during c2k10    found by bentley@ in sbcl(1)  Mon, 9 Dec 2013 18:36:57 -0700
   
 - \*(Pu is not implemented, but used in mdoc.samples  - .ns (no-space mode) occurs in xine-config(1)
   implementing .ds would solve this as well    reported by brad@  Sat, 15 Jan 2011 15:45:23 -0500
   reminded by jmc@, last on Sun, 27 Jun 2010 15:29:18 +0100  
   
   - .ta (tab settings) occurs in ircbug(1) and probably gnats(1)
     reported by brad@  Sat, 15 Jan 2011 15:50:51 -0500
     also Tcl_NewStringObj(3) via wiz@  Wed, 5 Mar 2014 22:27:43 +0100
   
   - .ti (temporary indent)
     found by naddy@ in xloadimage(1)
     found by bentley@ in nmh(1)  Mon, 23 Apr 2012 13:38:28 -0600
   
   - .while and .shift
     found by jca@ in ratpoison(1)  Sun, 30 Jun 2013 12:01:09 +0200
   
   - \c (interrupted text) should prevent the line break
     even inside .Bd literal; that occurs in chat(8)
     also found in cclive(1) - DocBook output
   
   - \h horizontal move
     found in cclive(1) DocBook output
     Anthony J. Bentley on discuss@  Sat, 21 Sep 2013 22:29:34 -0600
   
   - \n+ and \n- numerical register increment and decrement
     found by bentley@ in sbcl(1)  Mon, 9 Dec 2013 18:36:57 -0700
   
   - \w'' width measurements
     would not be very useful without an expression parser, see below
     needed for Tcl_NewStringObj(3) via wiz@  Wed, 5 Mar 2014 22:27:43 +0100
   
   - using undefined strings or macros defines them to be empty
     wl@  Mon, 14 Nov 2011 14:37:01 +0000
   
   - general expression parser, including arithmetics
     to be used at least for .if/.ie and .nr and maybe at other places
     could use J.T.Conklin's PD code in bin/expr/expr.c for inspiration
     needed for Tcl_NewStringObj(3) via wiz@  Wed, 5 Mar 2014 22:27:43 +0100
   
   --- missing mdoc features ----------------------------------------------
   
   - fix bad block nesting involving multiple identical explicit blocks
     see the OpenBSD mdoc_macro.c 1.47 commit message
   
 - .Bl -column .Xo support is missing  - .Bl -column .Xo support is missing
   ultimate goal:    ultimate goal:
   restore .Xr and .Dv to    restore .Xr and .Dv to
Line 26 
Line 81 
   lib/libc/gen/signal.3    lib/libc/gen/signal.3
   lib/libc/sys/sigaction.2    lib/libc/sys/sigaction.2
   
 - implement \\  - edge case: decide how to deal with blk_full bad nesting, e.g.
   in plain text, identical to \e    .Sh .Nm .Bk .Nm .Ek .Sh found by jmc@ in ssh-keygen(1)
   as a macro argument, identical to \ i.e. escaping the next character    from jmc@  Wed, 14 Jul 2010 18:10:32 +0100
   We do not have macro definitions yet; if we implement them,  
   \\ must behave in a macro def like in a macro argument,  
   and when using the macro, it must expand yet again.  
   
 - look at bsd.lv tbl(1)  - \\ is now implemented correctly
   from kristaps@  Fri, 11 Sep 2009 17:10:53 +0200    * when defining strings and macros using .ds and .de
   also look at the mail from Thomas Klausner wiz at NetBSD    * when parsing roff(7) and man(7) macro arguments
     on Wed, 2 Jun 2010 11:01:29 +0200    It does not yet work in mdoc(7) macro arguments
     because libmdoc does not yet use mandoc_getarg().
     Also check what happens in plain text, it must be identical to \e.
   
 - look at pages generated from reStructeredText, e.g. devel/mercurial hg(1)  - .Bd -filled should not be the same as .Bd -ragged, but align both
   These are a weird mixture of man(7) and custom autogenerated low-level    the left and right margin.  In groff, it is implemented in terms
   roff stuff.  Figure out to what extent we can cope.    of .ad b, which we don't have either.  Found in cksum(1).
   noted by stsp@  Sat, 24 Apr 2010 09:17:55 +0200  
   reminded by nicm@  Mon, 3 May 2010 09:52:41 +0100  
   
 - implement blank `Bl -column', such as  - implement blank `Bl -column', such as
   .Bl -column    .Bl -column
Line 52 
Line 104 
 - explicitly disallow nested `Bl -column', which would clobber internal  - explicitly disallow nested `Bl -column', which would clobber internal
   flags defined for struct mdoc_macro    flags defined for struct mdoc_macro
   
   - In .Bl -column .It, the end of the line probably has to be regarded
     as an implicit .Ta, if there could be one, see the following mildly
     ugly code from login.conf(5):
       .Bl -column minpasswordlen program xetcxmotd
       .It path Ta path Ta value of Dv _PATH_DEFPATH
       .br
       Default search path.
     reported by Michal Mazurek <akfaew at jasminek dot net>
     via jmc@ Thu, 7 Apr 2011 16:00:53 +0059
   
   - inside `.Bl -column' phrases, punctuation is handled like normal
     text, e.g. `.Bl -column .It Fl x . Ta ...' should give "-x -."
   
   - inside `.Bl -column' phrases, TERMP_IGNDELIM handling by `Pf'
     is not safe, e.g. `.Bl -column .It Pf a b .' gives "ab."
     but should give "ab ."
   
 - set a meaningful default if no `Bl' list type is assigned  - set a meaningful default if no `Bl' list type is assigned
   
 - have a blank `It' head for `Bl -tag' not puke  - have a blank `It' head for `Bl -tag' not puke
   
 - ignore horrendous m[] font colouring  
   reported by J.C. Roberts  
   
 - prohibit `Nm' from having non-text HEAD children  - prohibit `Nm' from having non-text HEAD children
   (e.g., NetBSD mDNSShared/dns-sd.1)    (e.g., NetBSD mDNSShared/dns-sd.1)
   (mdoc_html.c and mdoc_term.c `Nm' handlers can be slightly simplified)    (mdoc_html.c and mdoc_term.c `Nm' handlers can be slightly simplified)
   
   - When there is free text in the SYNOPSIS and that free text contains
     the .Nm macro, groff somehow understands to treat the .Nm as an in-line
     macro, while mandoc treats it as a block macro and breaks the line.
     No idea how the logic for distinguishing in-line and block instances
     should be, needs investigation.
     uqs@  Thu, 2 Jun 2011 11:03:51 +0200
     uqs@  Thu, 2 Jun 2011 11:33:35 +0200
   
   --- missing man features -----------------------------------------------
   
   - -T[x]html doesn't stipulate non-collapsing spaces in literal mode
   
   --- missing tbl features -----------------------------------------------
   
   - implement basic non-parametric .de to support e.g. sox(1)
     reported by naddy@ Sat, 16 Oct 2010 23:51:57 +0200
     *** sox(1) still doesn't work, tbl(1) errors need investigation
   
   - allow standalone `.' to be interpreted as an end-of-layout
     delimiter instead of being thrown away as a no-op roff line
     reported by Yuri Pankov, Wed 18 May 2011 11:34:59 CEST
   
   --- missing misc features ----------------------------------------------
   
   - italic correction (\/) in PostScript mode
     Werner LEMBERG on groff at gnu dot org  Sun, 10 Nov 2013 12:47:46
   
   - The whatis(1) utility looks for whole words in Nm.
     If the file name of a page does not agree with the contents of any
     of its Nm macros (e.g. pool(9)), add the file name as an Nm entry
     to the mandoc.db as well, such that whatis(1) finds it.
     If there is a page with a file name that does not appear as a substring
     neither in Nm nor in Nd, the same fix would allow finding that page
     with apropos(1) using the file name as a key, as well.
     Issue reported by tedu@  Fri, 05 Jul 2013 21:15:23 -0400
   
   - clean up escape sequence handling, creating three classes:
     (1) fully implemented, or parsed and ignored without loss of content
     (2) unimplemented, potentially causing loss of content
         or serious mangling of formatting (e.g. \n) -> ERROR
         see textproc/mgdiff(1) for nice examples
     (3) undefined, just output the character -> perhaps WARNING
   
   - look at pages generated from reStructeredText, e.g. devel/mercurial hg(1)
     These are a weird mixture of man(7) and custom autogenerated low-level
     roff stuff.  Figure out to what extent we can cope.
     For details, see http://docutils.sourceforge.net/rst.html
     noted by stsp@  Sat, 24 Apr 2010 09:17:55 +0200
     reminded by nicm@  Mon, 3 May 2010 09:52:41 +0100
   
   - look at pages generated from Texinfo source by yat2m, e.g. security/gnupg
     First impression is not that bad.
   
   - check compatibility with Plan9:
     http://swtch.com/usr/local/plan9/tmac/tmac.an
     http://swtch.com/plan9port/man/man7/man.html
     "Anthony J. Bentley" <anthonyjbentley@gmail.com> 28 Dec 2010 21:58:40 -0700
   
 ************************************************************************  ************************************************************************
 * formatting issues: ugly output  * formatting issues: ugly output
 ************************************************************************  ************************************************************************
   
 - outside list context, text following .Sm off  - a column list with blank `Ta' cells triggers a spurrious
   seems to follow without a blank (TERM_NOBLANK reset late?) e.g.    start-with-whitespace printing of a newline
         .Sm off  
         .Xo  
         .Op Ar bind_address No /  
         .Ar port  
         .Xc  
         .Sm on  
         or by enclosing  
   reported by jmc Tue, 13 Apr 2010 08:48:14 +0100  
   
 - indentation got lost in SYNOPSIS, e.g.  - In .Bl -column,
    SYNOPSIS    .It Em Authentication<tab>Key Length
         date [-ajnu] [-d dst] [-r seconds] [-t minutes_west] [+format]    ought to render "Key Length" with emphasis, too,
   -          [[[[[[cc]yy]mm]dd]HH]MM[.SS]]    see OpenBSD iked.conf(5).
   +     [[[[[[cc]yy]mm]dd]HH]MM[.SS]]    reported again Nicolas Joly via wiz@ Wed, 12 Oct 2011 00:20:00 +0200
   reminded by Ludo Smissaert  Mon, 10 May 2010 12:15:11 +0200  
   
 - empty phrases in .Bl column produce too few blanks  - empty phrases in .Bl column produce too few blanks
   try e.g. .Bl -column It Ta Ta    try e.g. .Bl -column It Ta Ta
   reported by millert Fri, 02 Apr 2010 16:13:46 -0400    reported by millert Fri, 02 Apr 2010 16:13:46 -0400
   
   - .%T can have trailing punctuation.  Currently, it puts the trailing
     punctuation into a trailing MDOC_TEXT element inside its own scope.
     That element should rather be outside its scope, such that the
     punctuation does not get underlines.  This is not trivial to
     implement because .%T then needs some features of in_line_eoln() -
     slurp all arguments into one single text element - and one feature
     of in_line() - put trailing punctuation out of scope.
     Found in mount_nfs(8) and exports(5), search for "Appendix".
   
   - Trailing punctuation after .%T triggers EOS spacing, at least
     outside .Rs (eek!).  Simply setting ARGSFL_DELIM for .%T is not
     the right solution, it sends mandoc into an endless loop.
     reported by Nicolas Joly  Sat, 17 Nov 2012 11:49:54 +0100
   
   - in enclosures, mandoc sometimes fancies a bogus end of sentence
     reminded by jmc@  Thu, 23 Sep 2010 18:13:39 +0059
   
   - formatting /usr/local/man/man1/latex2man.1 with groff and mandoc
     reveals lots of bugs both in groff and mandoc...
     reported by bentley@  Wed, 22 May 2013 23:49:30 -0600
   
 ************************************************************************  ************************************************************************
 * formatting issues: gratuitious differences  * formatting issues: gratuitous differences
 ************************************************************************  ************************************************************************
   
 - none reported right now (yet schwarze@ knows of some)  - .Rv (and probably .Ex) print different text if an `Nm' has been named
     or not (run a manual without `Nm blah' to see this).  I'm not sure
     that this exists in the wild, but it's still an error.
   
   - In .Bl -bullet, the groff bullet is "+\b+\bo\bo", the mandoc bullet
     is just "o\bo".
     see for example OpenBSD ksh(1)
   
   - .Pp between two .It in .Bl -column should produce one,
     not two blank lines, see e.g. login.conf(5).
     reported by jmc@  Sun, 17 Apr 2011 14:04:58 +0059
     reported again by sthen@  Wed, 18 Jan 2012 02:09:39 +0000 (UTC)
   
   - If the *first* line after .It is .Pp, break the line right after
     the tag, do not pad with space characters before breaking.
     See the description of the a, c, and i commands in sed(1).
   
   - If the first line after .It is .D1, do not assert a blank line
     in between, see for example tmux(1).
     reported by nicm@  13 Jan 2011 00:18:57 +0000
   
   - Trailing punctuation after .It should trigger EOS spacing.
     reported by Nicolas Joly  Sat, 17 Nov 2012 11:49:54 +0100
     Probably, this should be fixed somewhere in termp_it_pre(), not sure.
   
   - .Nx 1.0a
     should be "NetBSD 1.0A", not "NetBSD 1.0a",
     see OpenBSD ccdconfig(8).
   
   - In .Bl -tag, if a tag exceeds the right margin and must be continued
     on the next line, it must be indented by -width, not width+1;
     see "rule block|pass" in OpenBSD ifconfig(8).
   
   - When the -width string contains macros, the macros must be rendered
     before measuring the width, for example
       .Bl -tag -width ".Dv message"
     in magic(5), located in src/usr.bin/file, is the same
     as -width 7n, not -width 11n.
     The same applies to .Bl -column column widths;
     reported again by Nicolas Joly Thu, 1 Mar 2012 13:41:26 +0100 via wiz@ 5 Mar
     reported again by Franco Fichtner Fri, 27 Sep 2013 21:02:28 +0200
     An easy partial fix would be to just skip the first word if it starts
     with a dot, including any following white space, when measuring.
   
   - The \& zero-width character counts as output.
     That is, when it is alone on a line between two .Pp,
     we want three blank lines, not two as in mandoc.
   
   - Header lines of excessive length:
     Port OpenBSD man_term.c rev. 1.25 to mdoc_term.c
     and document it in mdoc(7) and man(7) COMPATIBILITY
     found while talking to Chris Bennett
   
   - trailing whitespace must be ignored even when followed by a font escape,
     see for example
       makes
       \fBdig \fR
       operate in batch mode
     in dig(1).
   
 ************************************************************************  ************************************************************************
   * warning issues
   ************************************************************************
   
   - check that MANDOCERR_BADTAB is thrown in the right cases,
     i.e. when finding a literal tab character in fill mode,
     and possibly change the wording of the warning message
     to refer to fill mode, not literal mode
     See the mail from Werner LEMBERG on the groff list,
     Fri, 14 Feb 2014 18:54:42 +0100 (CET)
   
   ************************************************************************
 * performance issues  * performance issues
 ************************************************************************  ************************************************************************
   
 Several areas can be cleaned up to make mandoc even faster.  These are  Several areas can be cleaned up to make mandoc even faster.  These are
   
 - improve caching mechanism to handle other per-element arguments so  
   that they need not be recalculated between front- and back-ends (also  
   reduces code size and complexity):  
     `Bl -column' pointer  
     `Bf' font type  
   
 - improve hashing mechanism for macros (quite important: performance)  - improve hashing mechanism for macros (quite important: performance)
   
 - improve hashing mechanism for characters (not as important)  - improve hashing mechanism for characters (not as important)
   
   - the PDF file is HUGE: this can be reduced by using relative offsets
   
   - instead of re-initialising the roff predefined-strings set before each
     parse, create a read-only version the first time and copy it
   
   ************************************************************************
   * structural issues
   ************************************************************************
   
   - We use the input line number at several places to distinguish
     same-line from different-line input.  That plainly doesn't work
     with user-defined macros, leading to random breakage.
   
   - Find better ways to prevent endless loops
     in roff(7) macro and string expansion.
   
   - Finish cleanup of date handling.
     Decide which formats should be recognized where.
     Update both mdoc(7) and man(7) documentation.
     Triggered by  Tim van der Molen  Tue, 22 Feb 2011 20:30:45 +0100

Legend:
Removed from v.1.20  
changed lines
  Added in v.1.169

CVSweb