From: Nathan Bossart <nathandbossart@gmail.com>
To: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Cc: michael@paquier.xyz
Cc: noah@leadboat.com
Cc: pgsql-hackers@postgresql.org
Subject: Re: add non-option reordering to in-tree getopt_long
Date: Mon, 10 Jul 2023 13:06:58 -0700
Message-ID: <20230710200658.GA362078@nathanxps13> (raw)
In-Reply-To: <20230710181237.GD257945@nathanxps13>
References: <20230620.141244.1061039848681375358.horikyota.ntt@gmail.com>
<20230621175624.GA882103@nathanxps13>
<20230708035224.GA4042213@nathanxps13>
<20230710.165711.1906730109995973216.horikyota.ntt@gmail.com>
<20230710181237.GD257945@nathanxps13>
Here's a new version of the patch with the latest feedback addressed.
--
Nathan Bossart
Amazon Web Services: https://aws.amazon.comAttachments:v6-0001-Teach-in-tree-getopt_long-to-move-non-options-to-.patch (text/x-diff, 5.2K) download
From e662a1b6b73c92ff7862444e092406ed82b0c2c7 Mon Sep 17 00:00:00 2001
From: Nathan Bossart <nathan@postgresql.org>
Date: Fri, 7 Jul 2023 20:00:47 -0700
Subject: [PATCH v6 1/1] Teach in-tree getopt_long() to move non-options to the
end of argv.
Unlike the other implementations of getopt_long() I could find, the
in-tree implementation does not reorder non-options to the end of
argv. Instead, it returns -1 as soon as the first non-option is
found, even if there are other options listed afterwards. By
moving non-options to the end of argv, getopt_long() can parse all
specified options and return -1 when only non-options remain.
This quirk is periodically missed by hackers (e.g., 869aa40a27,
ffd398021c, and d9ddc50baf).
This commit introduces the aforementioned non-option reordering
behavior to the in-tree getopt_long() implementation. The only
systems I'm aware of that use it are Windows and AIX, both of which
have been tested.
Special thanks to Noah Misch for his help validating behavior on
AIX.
Reviewed-by: Kyotaro Horiguchi, Michael Paquier
Discussion: https://postgr.es/m/20230609232257.GA121461%40nathanxps13
---
src/bin/scripts/t/040_createuser.pl | 10 +++---
src/port/getopt_long.c | 54 ++++++++++++++++++++---------
2 files changed, 43 insertions(+), 21 deletions(-)
diff --git a/src/bin/scripts/t/040_createuser.pl b/src/bin/scripts/t/040_createuser.plindex 9ca282181d..3290e30c88 100644--- a/src/bin/scripts/t/040_createuser.pl+++ b/src/bin/scripts/t/040_createuser.pl@@ -42,9 +42,8 @@ $node->issues_sql_like(
'add a role as a member with admin option of the newly created role');
$node->issues_sql_like(
[
- 'createuser', '-m',- 'regress_user3', '-m',- 'regress user #4', 'REGRESS_USER5'+ 'createuser', 'REGRESS_USER5', '-m', 'regress_user3',+ '-m', 'regress user #4'
],
qr/statement: CREATE ROLE "REGRESS_USER5" NOSUPERUSER NOCREATEDB NOCREATEROLE INHERIT LOGIN NOREPLICATION NOBYPASSRLS ROLE regress_user3,"regress user #4";/,
'add a role as a member of the newly created role');
@@ -73,11 +72,14 @@ $node->issues_sql_like(
qr/statement: CREATE ROLE regress_user11 NOSUPERUSER NOCREATEDB NOCREATEROLE INHERIT LOGIN NOREPLICATION NOBYPASSRLS IN ROLE regress_user1;/,
'--role');
$node->issues_sql_like(
- [ 'createuser', '--member-of', 'regress_user1', 'regress_user12' ],+ [ 'createuser', 'regress_user12', '--member-of', 'regress_user1' ],
qr/statement: CREATE ROLE regress_user12 NOSUPERUSER NOCREATEDB NOCREATEROLE INHERIT LOGIN NOREPLICATION NOBYPASSRLS IN ROLE regress_user1;/,
'--member-of');
$node->command_fails([ 'createuser', 'regress_user1' ],
'fails if role already exists');
+$node->command_fails(+ [ 'createuser', 'regress_user1', '-m', 'regress_user2', 'regress_user3' ],+ 'fails for too many non-options');
done_testing();
diff --git a/src/port/getopt_long.c b/src/port/getopt_long.cindex c989276988..3ba6094d93 100644--- a/src/port/getopt_long.c+++ b/src/port/getopt_long.c@@ -50,8 +50,9 @@
* This implementation does not use optreset. Instead, we guarantee that
* it can be restarted on a new argv array after a previous call returned -1,
* if the caller resets optind to 1 before the first call of the new series.
- * (Internally, this means we must be sure to reset "place" to EMSG before- * returning -1.)+ *+ * Note that this routine reorders the pointers in argv (despite the const+ * qualifier) so that all non-options will be at the end when -1 is returned.
*/
int
getopt_long(int argc, char *const argv[],
@@ -60,40 +61,59 @@ getopt_long(int argc, char *const argv[],
{
static char *place = EMSG; /* option letter processing */
char *oli; /* option letter list index */
+ static int nonopt_start = -1;+ static bool force_nonopt = false;
if (!*place)
{ /* update scanning pointer */
- if (optind >= argc)+ char **args = (char **) argv;++retry:++ /*+ * If we are out of arguments or only non-options remain, return -1.+ */+ if (optind >= argc || optind == nonopt_start)
{
place = EMSG;
+ nonopt_start = -1;+ force_nonopt = false;
return -1;
}
place = argv[optind];
- if (place[0] != '-')+ /*+ * An argument is a non-option if it meets any of the following+ * criteria: it follows an argument that is equivalent to the string+ * "--", it is equivalent to the string "-", or it does not start with+ * '-'. When we encounter a non-option, we move it to the end of argv+ * (after shifting all remaining arguments over to make room), and+ * then we try again with the next argument.+ */+ if (force_nonopt || strcmp("-", place) == 0 || place[0] != '-')
{
- place = EMSG;- return -1;- }+ for (int i = optind; i < argc - 1; i++)+ args[i] = args[i + 1];+ args[argc - 1] = place;- place++;+ if (nonopt_start == -1)+ nonopt_start = argc - 1;+ else+ nonopt_start--;- if (!*place)- {- /* treat "-" as not being an option */- place = EMSG;- return -1;+ goto retry;
}
-- if (place[0] == '-' && place[1] == '\0')+ else if (strcmp("--", place) == 0)
{
/* found "--", treat it as end of options */
++optind;
- place = EMSG;- return -1;+ force_nonopt = true;+ goto retry;
}
+ place++;+
if (place[0] == '-' && place[1])
{
/* long option */
--
2.25.1
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients from your mail client; the link fills in
the headers below: reply via email
Some clients, webmail especially, drop In-Reply-To and References from
a link, and the reply then starts a new thread: use the mbox below.
To: pgsql-hackers@postgresql.org
Cc: nathandbossart@gmail.com, horikyota.ntt@gmail.com, michael@paquier.xyz, noah@leadboat.com
Subject: Re: add non-option reordering to in-tree getopt_long
In-Reply-To: <20230710200658.GA362078@nathanxps13>
References: <20230620.141244.1061039848681375358.horikyota.ntt@gmail.com> <20230621175624.GA882103@nathanxps13> <20230708035224.GA4042213@nathanxps13> <20230710.165711.1906730109995973216.horikyota.ntt@gmail.com> <20230710181237.GD257945@nathanxps13> <20230710200658.GA362078@nathanxps13>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox