post-p2k11 resync

post-p2k11 resync

[Ingo Schwarze]

Hi,

the OpenBSD part of the resync is done (well, there are still a few
minor patches outside the mandocsb/apropos area, i'll resend those
later).

Here is the bsd.lv part, mostly removing trivial differences and
cleaning up whitespace.  But there are a few real issues:

 * -M overrides MANPATH, not the other way round;
   -m _prepends_ to the path, it doesn't _append_.
   That was both implemented and documented the wrong way.

 * MANPATH is colon-separated, not comma-separated;
   the implementation is correct, the manual in part wrong.

 * apropos.c doesn't need ctype.h any longer.

 * The apropos(1) usage() has -M and -m "path", not "dirs",
   just type "apropos -x" on an OpenBSD system.

 * I fear removing the database close calls from the dir loop
   in mandocdb(8) main because of file descriptor leaks;
   or am i missing something?

 * manpath.c needs sys/types.h because of size_t.

OK?
  Ingo


Index: apropos.1
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/apropos.1,v
retrieving revision 1.8
diff -u -p -r1.8 apropos.1
--- apropos.1	23 Nov 2011 10:09:30 -0000	1.8
+++ apropos.1	26 Nov 2011 16:45:13 -0000
@@ -44,7 +44,7 @@ searched for
 databases.
 Invalid paths, or paths without manual databases, are ignored.
 .It Fl m Ar manpath
-Append the colon-separated paths to the list of paths searched
+Prepend the colon-separated paths to the list of paths searched
 for
 .Xr mandocdb 8
 databases.
@@ -154,10 +154,10 @@ If an architecture is specified in the o
 .Sh ENVIRONMENT
 .Bl -tag -width Ds
 .It Ev MANPATH
-Comma-separated paths overriding the default list of paths searched for
+Colon-separated paths overriding the default list of paths searched for
 manual databases.
 Invalid paths, or paths without manual databases, are ignored.
-Takes precedence over
+Overridden by
 .Fl M .
 .El
 .\" .Sh FILES
Index: apropos.c
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/apropos.c,v
retrieving revision 1.18
diff -u -p -r1.18 apropos.c
--- apropos.c	23 Nov 2011 09:50:40 -0000	1.18
+++ apropos.c	26 Nov 2011 16:45:13 -0000
@@ -20,7 +20,6 @@
 #endif
 
 #include <assert.h>
-#include <ctype.h>
 #include <getopt.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -61,7 +60,7 @@ main(int argc, char *argv[])
 	e = NULL;
 	rc = 0;
 
-	while (-1 != (ch = getopt(argc, argv, "M:m:S:s:"))) 
+	while (-1 != (ch = getopt(argc, argv, "M:m:S:s:")))
 		switch (ch) {
 		case ('M'):
 			defpaths = optarg;
@@ -96,10 +95,10 @@ main(int argc, char *argv[])
 	}
 
 	rc = apropos_search
-		(paths.sz, paths.paths, 
+		(paths.sz, paths.paths,
 		 &opts, e, terms, NULL, list);
 
-	if (0 == rc) 
+	if (0 == rc)
 		fprintf(stderr, "%s: Error reading "
 				"manual database\n", progname);
 
@@ -119,8 +118,8 @@ list(struct res *res, size_t sz, void *a
 	qsort(res, sz, sizeof(struct res), cmp);
 
 	for (i = 0; i < (int)sz; i++)
-		printf("%s(%s%s%s) - %s\n", res[i].title, 
-				res[i].cat, 
+		printf("%s(%s%s%s) - %s\n", res[i].title,
+				res[i].cat,
 				*res[i].arch ? "/" : "",
 				*res[i].arch ? res[i].arch : "",
 				res[i].desc);
@@ -139,9 +138,10 @@ usage(void)
 {
 
 	fprintf(stderr, "usage: %s "
-			"[-M dirs] "
-			"[-m dirs] "
+			"[-M path] "
+			"[-m path] "
 			"[-S arch] "
 			"[-s section] "
-			"expression...\n", progname);
+			"expression...\n",
+			progname);
 }
Index: apropos_db.c
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/apropos_db.c,v
retrieving revision 1.11
diff -u -p -r1.11 apropos_db.c
--- apropos_db.c	23 Nov 2011 09:55:28 -0000	1.11
+++ apropos_db.c	26 Nov 2011 16:45:13 -0000
@@ -115,12 +115,12 @@ static	const struct type types[] = {
 };
 
 static	DB	*btree_open(void);
-static	int	 btree_read(const DBT *, 
+static	int	 btree_read(const DBT *,
 			const struct mchars *, char **);
 static	int	 expreval(const struct expr *, int *);
-static	void	 exprexec(const struct expr *, 
+static	void	 exprexec(const struct expr *,
 			const char *, uint64_t, struct rec *);
-static	int	 exprmark(const struct expr *, 
+static	int	 exprmark(const struct expr *,
 			const char *, uint64_t, int *);
 static	struct expr *exprexpr(int, char *[], int *, int *, size_t *);
 static	struct expr *exprterm(char *, int);
@@ -148,7 +148,7 @@ btree_open(void)
 	info.flags = R_DUP;
 
 	db = dbopen(MANDOC_DB, O_RDONLY, 0, DB_BTREE, &info);
-	if (NULL != db) 
+	if (NULL != db)
 		return(db);
 
 	return(NULL);
@@ -176,7 +176,7 @@ btree_read(const DBT *v, const struct mc
 /*
  * Take a Unicode codepoint and produce its UTF-8 encoding.
  * This isn't the best way to do this, but it works.
- * The magic numbers are from the UTF-8 packaging.  
+ * The magic numbers are from the UTF-8 packaging.
  * They're not as scary as they seem: read the UTF-8 spec for details.
  */
 static size_t
@@ -241,7 +241,7 @@ norm_string(const char *val, const struc
 	const char	 *seq, *cpp;
 	int		  len, u, pos;
 	enum mandoc_esc	  esc;
-	static const char res[] = { '\\', '\t', 
+	static const char res[] = { '\\', '\t',
 				ASCII_NBRSP, ASCII_HYPH, '\0' };
 
 	/* Pre-allocate by the length of the input */
@@ -287,7 +287,7 @@ norm_string(const char *val, const struc
 		if (ESCAPE_ERROR == esc)
 			break;
 
-		/* 
+		/*
 		 * XXX - this just does UTF-8, but we need to know
 		 * beforehand whether we should do text substitution.
 		 */
@@ -382,7 +382,7 @@ index_read(const DBT *key, const DBT *va
  */
 int
 apropos_search(int pathsz, char **paths, const struct opts *opts,
-		const struct expr *expr, size_t terms, void *arg, 
+		const struct expr *expr, size_t terms, void *arg,
 		void (*res)(struct res *, size_t, void *))
 {
 	struct rectree	 tree;
@@ -420,7 +420,7 @@ apropos_search(int pathsz, char **paths,
 
 	for (mlen = i = 0; i < tree.len; i++)
 		if (tree.node[i].matched)
-			memcpy(&ress[mlen++], &tree.node[i].res, 
+			memcpy(&ress[mlen++], &tree.node[i].res,
 					sizeof(struct res));
 
 	(*res)(ress, mlen, arg);
@@ -460,7 +460,7 @@ single_search(struct rectree *tree, cons
 
 	memset(&r, 0, sizeof(struct rec));
 
-	if (NULL == (btree = btree_open())) 
+	if (NULL == (btree = btree_open()))
 		return(1);
 
 	if (NULL == (idx = index_open())) {
@@ -469,8 +469,8 @@ single_search(struct rectree *tree, cons
 	}
 
 	while (0 == (ch = (*btree->seq)(btree, &key, &val, R_NEXT))) {
- 		if (key.size < 2 || sizeof(struct db_val) != val.size) 
-  			break;
+		if (key.size < 2 || sizeof(struct db_val) != val.size)
+			break;
 		if ( ! btree_read(&key, mc, &buf))
 			break;
 
@@ -492,13 +492,13 @@ single_search(struct rectree *tree, cons
 		 */
 
 		for (leaf = root; leaf >= 0; )
-			if (rec > rs[leaf].res.rec && 
+			if (rec > rs[leaf].res.rec &&
 					rs[leaf].rhs >= 0)
 				leaf = rs[leaf].rhs;
-			else if (rec < rs[leaf].res.rec && 
+			else if (rec < rs[leaf].res.rec &&
 					rs[leaf].lhs >= 0)
 				leaf = rs[leaf].lhs;
-			else 
+			else
 				break;
 
 		/*
@@ -540,10 +540,11 @@ single_search(struct rectree *tree, cons
 			(rs, (tree->len + 1) * sizeof(struct rec));
 
 		memcpy(&rs[tree->len], &r, sizeof(struct rec));
-		rs[tree->len].matches = 
+		rs[tree->len].matches =
 			mandoc_calloc(terms, sizeof(int));
 
-		exprexec(expr, buf, mask, &rs[tree->len]); 
+		exprexec(expr, buf, mask, &rs[tree->len]);
+
 		/* Append to our tree. */
 
 		if (leaf >= 0) {
@@ -553,11 +554,11 @@ single_search(struct rectree *tree, cons
 				rs[leaf].lhs = tree->len;
 		} else
 			root = tree->len;
-		
+
 		memset(&r, 0, sizeof(struct rec));
 		tree->len++;
 	}
-	
+
 	(*btree->close)(btree);
 	(*idx->close)(idx);
 
@@ -632,7 +633,7 @@ exprexpr(int argc, char *argv[], int *po
 		log = 0;
 
 		if (NULL != e && 0 == strcmp("-a", argv[*pos]))
-			log = 1;			
+			log = 1;
 		else if (NULL != e && 0 == strcmp("-o", argv[*pos]))
 			log = 2;
 
@@ -744,7 +745,7 @@ void
 exprfree(struct expr *p)
 {
 	struct expr	*pp;
-	
+
 	while (NULL != p) {
 		if (p->subexpr)
 			exprfree(p->subexpr);
@@ -758,7 +759,7 @@ exprfree(struct expr *p)
 }
 
 static int
-exprmark(const struct expr *p, const char *cp, 
+exprmark(const struct expr *p, const char *cp,
 		uint64_t mask, int *ms)
 {
 
@@ -813,7 +814,7 @@ expreval(const struct expr *p, int *ms)
 		for ( ; p->next && p->next->and; p = p->next) {
 			/* Evaluate a subexpression, if applicable. */
 			if (p->next->subexpr && ! ms[p->next->index])
-				ms[p->next->index] = 
+				ms[p->next->index] =
 					expreval(p->next->subexpr, ms);
 			match = match && ms[p->next->index];
 		}
@@ -830,11 +831,11 @@ expreval(const struct expr *p, int *ms)
  * If this evaluates to true, mark the expression as satisfied.
  */
 static void
-exprexec(const struct expr *p, const char *cp, 
+exprexec(const struct expr *e, const char *cp,
 		uint64_t mask, struct rec *r)
 {
 
 	assert(0 == r->matched);
-	exprmark(p, cp, mask, r->matches);
-	r->matched = expreval(p, r->matches);
+	exprmark(e, cp, mask, r->matches);
+	r->matched = expreval(e, r->matches);
 }
Index: apropos_db.h
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/apropos_db.h,v
retrieving revision 1.7
diff -u -p -r1.7 apropos_db.h
--- apropos_db.h	23 Nov 2011 09:55:28 -0000	1.7
+++ apropos_db.h	26 Nov 2011 16:45:13 -0000
@@ -24,11 +24,11 @@ struct	res {
 	char		*arch; /* arch (or empty string) */
 	char		*desc; /* description (from Nd) */
 	unsigned int	 rec; /* record in index */
-	/* 
+	/*
 	 * The index volume.  This indexes into the array of directories
 	 * searched for manual page databases.
 	 */
-	unsigned int	 volume; 
+	unsigned int	 volume;
 };
 
 struct	opts {
@@ -40,8 +40,8 @@ __BEGIN_DECLS
 
 struct	expr;
 
-int	 	 apropos_search(int, char **, const struct opts *, 
-			const struct expr *, size_t, void *, 
+int	 	 apropos_search(int, char **, const struct opts *,
+			const struct expr *, size_t, void *,
 			void (*)(struct res *, size_t, void *));
 struct	expr	*exprcomp(int, char *[], size_t *);
 void		 exprfree(struct expr *);
Index: mandocdb.c
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/mandocdb.c,v
retrieving revision 1.12
diff -u -p -r1.12 mandocdb.c
--- mandocdb.c	26 Nov 2011 11:23:56 -0000	1.12
+++ mandocdb.c	26 Nov 2011 16:45:13 -0000
@@ -378,7 +378,7 @@ main(int argc, char *argv[])
 				&maxrec, &recs, &recsz);
 
 		if (OP_UPDATE == op)
-			index_merge(of, mp, &dbuf, &buf, hash, 
+			index_merge(of, mp, &dbuf, &buf, hash,
 					db, fbuf, idx, ibuf, use_all,
 					verb, maxrec, recs, reccur);
 
@@ -411,11 +411,16 @@ main(int argc, char *argv[])
 		sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
 
 		if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
-			fprintf(stderr, "%s: Path too long\n", 
+			fprintf(stderr, "%s: Path too long\n",
 					dirs.paths[i]);
 			exit((int)MANDOCLEVEL_BADARG);
 		}
 
+		if (db)
+			(*db->close)(db);
+		if (idx)
+			(*idx->close)(idx);
+
 		db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
 		idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
 
@@ -444,7 +449,7 @@ main(int argc, char *argv[])
 
 		of = of->first;
 
-		index_merge(of, mp, &dbuf, &buf, hash, db, fbuf, 
+		index_merge(of, mp, &dbuf, &buf, hash, db, fbuf,
 				idx, ibuf, use_all, verb,
 				maxrec, recs, reccur);
 	}
@@ -471,7 +476,7 @@ out:
 void
 index_merge(const struct of *of, struct mparse *mp,
 		struct buf *dbuf, struct buf *buf,
-		DB *hash, DB *db, const char *dbf, 
+		DB *hash, DB *db, const char *dbf,
 		DB *idx, const char *idxf, int use_all, int verb,
 		recno_t maxrec, const recno_t *recs, size_t reccur)
 {
Index: manpath.c
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/manpath.c,v
retrieving revision 1.3
diff -u -p -r1.3 manpath.c
--- manpath.c	24 Nov 2011 10:44:56 -0000	1.3
+++ manpath.c	26 Nov 2011 16:45:13 -0000
@@ -19,6 +19,7 @@
 #include "config.h"
 #endif
 
+#include <sys/types.h>
 #include <assert.h>
 #include <ctype.h>
 #include <limits.h>
@@ -35,25 +36,25 @@
 static	void	 manpath_add(struct manpaths *, const char *);
 
 void
-manpath_parse(struct manpaths *dirs, char *defp, char *auxp) 
+manpath_parse(struct manpaths *dirs, char *defp, char *auxp)
 {
 
-	if (NULL != getenv("MANPATH"))
+	manpath_parseline(dirs, auxp);
+
+	if (NULL == defp)
 		defp = getenv("MANPATH");
 
 	if (NULL == defp)
 		manpath_parseconf(dirs);
 	else
 		manpath_parseline(dirs, defp);
-
-	manpath_parseline(dirs, auxp);
 }
 
 /*
  * Parse a FULL pathname from a colon-separated list of arrays.
  */
 void
-manpath_parseline(struct manpaths *dirs, char *path) 
+manpath_parseline(struct manpaths *dirs, char *path)
 {
 	char	*dir;
 
@@ -69,7 +70,7 @@ manpath_parseline(struct manpaths *dirs,
  * Grow the array one-by-one for simplicity's sake.
  */
 static void
-manpath_add(struct manpaths *dirs, const char *dir) 
+manpath_add(struct manpaths *dirs, const char *dir)
 {
 	char		 buf[PATH_MAX];
 	char		*cp;
@@ -83,7 +84,7 @@ manpath_add(struct manpaths *dirs, const
 			return;
 
 	dirs->paths = mandoc_realloc
-		(dirs->paths, 
+		(dirs->paths,
 		 ((size_t)dirs->sz + 1) * sizeof(char *));
 
 	dirs->paths[dirs->sz++] = mandoc_strdup(cp);
@@ -143,7 +144,7 @@ manpath_manconf(const char *file, struct
 {
 	FILE		*stream;
 	char		*p, *q;
-	size_t	 	 len, keysz;
+	size_t		 len, keysz;
 
 	keysz = strlen(MAN_CONF_KEY);
 	assert(keysz > 0);
--
 To unsubscribe send an email to tech+unsubscribe@mdocml.bsd.lv

Re: post-p2k11 resync

[Kristaps Dzonsons]

On 26/11/2011 17:57, Ingo Schwarze wrote:
> Hi,
>
> the OpenBSD part of the resync is done (well, there are still a few
> minor patches outside the mandocsb/apropos area, i'll resend those
> later).
>
> Here is the bsd.lv part, mostly removing trivial differences and
> cleaning up whitespace.  But there are a few real issues:
>
>   * -M overrides MANPATH, not the other way round;
>     -m _prepends_ to the path, it doesn't _append_.
>     That was both implemented and documented the wrong way.
>
>   * MANPATH is colon-separated, not comma-separated;
>     the implementation is correct, the manual in part wrong.
>
>   * apropos.c doesn't need ctype.h any longer.
>
>   * The apropos(1) usage() has -M and -m "path", not "dirs",
>     just type "apropos -x" on an OpenBSD system.
>
>   * I fear removing the database close calls from the dir loop
>     in mandocdb(8) main because of file descriptor leaks;
>     or am i missing something?
>
>   * manpath.c needs sys/types.h because of size_t.

Ingo,

Great---ok kristaps!  As you can see, I tried to shove all of the 
MANPATH handling into manpath_parse(), which should set everything up 
from MANPATH, -m, -M, and man.conf/manpath(1) (which, yes, is extremely 
ugly).  Only man.cgi doesn't use manpath_parse() because it reads 
manup's configuration file.  By the way, catman is fine by me as a name; 
I didn't even know it existed.  I'll look more into it later.

Incidentally, I still think there's a case for MANPATH acting like -m as 
mentioned before.  How else does one maintain a private manual tree? 
Using "-m" each time man(1) is invoked is ugly.

Best,

Kristaps

--
 To unsubscribe send an email to tech+unsubscribe@mdocml.bsd.lv

Re: post-p2k11 resync

[Ingo Schwarze]

Hi Kristaps,

Kristaps Dzonsons wrote on Sat, Nov 26, 2011 at 07:27:10PM +0100:
> On 26/11/2011 17:57, Ingo Schwarze wrote:

>> the OpenBSD part of the resync is done (well, there are still a few
>> minor patches outside the mandocsb/apropos area, i'll resend those
>> later).
>>
>> Here is the bsd.lv part, mostly removing trivial differences and
>> cleaning up whitespace.  But there are a few real issues:
>>
>>  * -M overrides MANPATH, not the other way round;
>>    -m _prepends_ to the path, it doesn't _append_.
>>    That was both implemented and documented the wrong way.
[...]

> Great---ok kristaps!

Thanks, done.

> As you can see, I tried to shove all of the
> MANPATH handling into manpath_parse(), which should set everything
> up from MANPATH, -m, -M, and man.conf/manpath(1)

Yes, i like that approach.

> (which, yes, is extremely ugly).

Sure, but maybe you are right that the portable version on
bsd.lv needs it; i didn't merge the #ifdef's to OpenBSD, though.
We are not getting manpath(1) anytime soon.  ;-)

[...]
> Incidentally, I still think there's a case for MANPATH acting like
> -m as mentioned before.  How else does one maintain a private manual
> tree? Using "-m" each time man(1) is invoked is ugly.

Well, i see your point there, though i'm not sure how important
it is.  If we think that we need this kind of functionality, we
might consider implementing the following approach found
in Linux manpath(1):

  MANPATH
    If $MANPATH is set,  manpath  displays  its  value  rather  than
    determining  it on the fly.  If $MANPATH is prefixed by a colon,
    then the value of the variable is appended to  the  list  deter-
    mined from the content of the configuration files.  If the colon
    comes at the end of the value in the variable, then  the  deter-
    mined  list  is appended to the content of the variable.  If the
    value of the variable contains a double  colon  (::),  then  the
    determined  list is inserted in the middle of the value, between
    the two colons.

But i'd at least recommend to wait some time until the dust has
settled, basic functionality is working and integrated and consider
fancy new features after that.

Until then, for people desperately needing this, a simple

  alias man='man -m "/my/very/private/manpath"'

might do the trick.

Yours,
  Ingo
--
 To unsubscribe send an email to tech+unsubscribe@mdocml.bsd.lv