Re: [PATCH] Fix typesetting in Bugs section of 'git-rebase' man page (web version)
Hmmm. I've also had some problems as well getting a correct man page build although it's the non-web version. In my case I use the git- config.1 man page to detect the problem. Uninteresting lines have been replaced with "[...]" for brevity in the following samples. Using 'MANWIDTH=65 man -c ./git-config.1' the bad man page has a section that looks like this:
http.<url>.*
Any of the http.* options above can be applied
selectively to some urls. For a config key to
match a URL, each element of the config key is
compared to that of the URL, in the following
order: 1. Scheme (e.g., https in
https://example.com/). This field must
match exactly between the config key and
the URL.
[...]
5. User name (e.g., user in
https://user@example.com/repo.git). If the
[...]
precedence than a config key with a user
name.
The list above is ordered by decreasing
precedence; a URL that matches a config
[...]
be preferred over a config key match of
https://user@example.com. All URLs are normalized before attempting
any matching (the password part, if
[...]
any URLs visited as a result of a
redirection do not participate in matching. i18n.commitEncoding
Character encoding the commit messages are
stored in; Git itself does not care per se,
The rest of the man page continues to be incorrectly indented too far, but I've omitted it for brevity.
The good output for the same section looks like this:
http.<url>.*
Any of the http.* options above can be applied
selectively to some urls. For a config key to
match a URL, each element of the config key is
compared to that of the URL, in the following
order: 1. Scheme (e.g., https in
https://example.com/). This field must
match exactly between the config key and
the URL.
[...]
5. User name (e.g., user in
https://user@example.com/repo.git). If the
[...]
precedence than a config key with a user
name. The list above is ordered by decreasing
precedence; a URL that matches a config key's
[...]
preferred over a config key match of
https://user@example.com. All URLs are normalized before attempting any
matching (the password part, if embedded in the
[...]
URLs visited as a result of a redirection do
not participate in matching. i18n.commitEncoding
Character encoding the commit messages are
stored in; Git itself does not care per se, butI have two different systems to build these pages on:
System A:
OS X + Mac Ports
asciidoc --version
asciidoc 8.6.6
xmlto --version
xmlto version 0.0.24
docbook-xsl is 1.75.2_0
produces bad git-config.1System B:
Ubuntu
asciidoc --version
asciidoc 8.5.2
xmlto --version
xmlto version 0.0.23
docbook-xsl is 1.75.2+dfsg-3
produces good git-config.1The git-config.1 file is built like so:
1. asciidoc produces git-config.xml from git-config.txt
2. xmlto produces git-config.1 from git-config.xml
System A:
git-config.txt: md5=c63cbafddb1bc3d6c38b6b2941f8cbce
git-config.xml: md5=99c0caa45ec02e8712c346e8842a7ce6
git-config.1: md5=812f8d180ec82c2786b95e986219a5d0
System B:
git-config.txt: md5=c63cbafddb1bc3d6c38b6b2941f8cbce
git-config.xml: md5=99c0caa45ec02e8712c346e8842a7ce6
git-config.1: md5=7298467bedd0225d5edefa8700bc9b2a
So actually the problem here is not the asciidoc step, it's actually the xmlto step.
And, if I make the following changes to the git-config.1 file produced on system A:
diff --git git-config.1 git-config-fix.1
--- git-config.1
+++ git-config-fix.1
@@ -3299,7 +3299,7 @@ user
in
https://user@example\&.com/repo\&.git)\&. If the config key
has[...]
.RE
-.RS 4
+.sp
The list above is ordered by decreasing precedence; a URL that
[...]
https://user@example\&.com/foo/bar
a config key match of
@@ -4460,7 +4460,7 @@ normal
all
\- Show also individual files in untracked directories\&.
.RE
-.RS 4
+.sp
If this variable is not specified, it defaults to
\fInormal\fR\&. This variable can be overridden with the [...]
\fBgit-status\fR(1)Then it will format properly again (and in fact the $googlecode/git- manpages-*.tar.gz files have the same correct ".sp" instead of ".RS 4" lines).
Trying xmlto versions 0.0.23 and 0.0.25 on System A still produce the same wrong output, so while it appears to be the xmlto step that's the problem, it's not xmlto itself.
After much gnashing of teeth and digging it turns out the problem is in the docbook-xsl sources. Debian includes a patch file "8530_manpages_lists_indentation_fix.dpatch" in [1]:
---- BEGIN PATCH ----
#! /bin/sh /usr/share/dpatch/dpatch-run
## 8530_manpages_lists_indentation_fix.dpatch by Daniel Leidert (dale)
<daniel.leidert@wgdd.de>
##
## All lines beginning with `## DP:' are a description of the patch.
## DP: Upstream fix from revision 8530 ( dleidert ):
## DP: Fix two indentation issues: In the first case there is no
corresponding
## DP: .RS macro (Debian #519438, sf.net 2793873). In the second
case an .RS
## DP: instead of the probably intended .sp leads to an indentation
bug
## DP: (Debian #527309, sf.net #2642139).
## DP:
## DP: <URL:http://docbook.svn.sourceforge.net/docbook/?rev=8530&view=rev
>
## DP: <URL:http://sourceforge.net/search/?type_of_search=artifact&group_id=21935&words=2642139
>
## DP: <URL:http://bugs.debian.org/519438>
## DP: <URL:http://bugs.debian.org/527309>
@DPATCH@ diff -urNad trunk~/manpages/lists.xsl trunk/manpages/lists.xsl --- trunk~/manpages/lists.xsl 2009-03-27 02:25:08.000000000 +0100 +++ trunk/manpages/lists.xsl 2009-11-04 00:54:35.000000000 +0100
Jason St. John
