[PATCH 0/2] small mtd-utils fixes

[PATCH 0/2] small mtd-utils fixes

[Brian Norris]

Hi,

These two fixes are somewhat trivial, although the second one fixes a
kind of regression that occurred since the last release.

Speaking of releases: when will we make a new release? After kernel 3.2
is final? Or perhaps a pre-3.2 and post-3.2 release? There are a lot of
important fixes in git right now, and I'd like to incorporate them
officially.

Also, note that the following commit was intended to be paired with the
commits in the "brian" branch:

    commit bf01f2960ba82468b1b25f00e044fd0c3ee0770a
    nandwrite: consolidate buffer usage

While it is functionally harmless, it does include an unnecessary
memcpy(). In fact, the patch really is useful only when paired with the
rest of the 'brian' branch, which is intended for kernels supporting
ioctl(MEMWRITE) - i.e., kernel version >= 3.2.

So I recommend that the next release either includes the 'brian' branch
or does not include commit bf01f2. Mike Frysinger and I discussed this
issue on the ML earlier and didn't really come up with a solution that
avoids the overhead both with and without MEMWRITE. I figure pre-3.2
kernel users will just have to put up with it eventually.

Thanks,
Brian

Brian Norris (2):
  mtd-utils: add jffs2reader to .gitignore
  mtdinfo: provide info when used without arguments

 .gitignore          |    1 +
 ubi-utils/mtdinfo.c |   14 ++++++++++----
 2 files changed, 11 insertions(+), 4 deletions(-)

-- 
1.7.5.4

[PATCH 1/2] mtd-utils: add jffs2reader to .gitignore

[Brian Norris]

Signed-off-by: Brian Norris <computersforpeace@gmail.com>
---
 .gitignore |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/.gitignore b/.gitignore
index 3a56d0d..d4771fb 100644
--- a/.gitignore
+++ b/.gitignore
@@ -30,6 +30,7 @@
 /ftl_check
 /ftl_format
 /jffs2dump
+/jffs2reader
 /mkfs.jffs2
 /mtd_debug
 /nanddump
-- 
1.7.5.4

[PATCH 2/2] mtdinfo: provide info when used without arguments

[Brian Norris]

If the user specifies neither a MTD argument nor the `-a' flag, we print
a cryptic message; i.e.,

    # mtdinfo
    libmtd: error!: cannot get information about "(null)"
            error 14 (Bad address)
    mtdinfo: error!: cannot get information about MTD device "(null)"
             error 14 (Bad address)

This is a regression; previously, mtdinfo would give some short info
about number of devices, etc. when used without arguments. To fix this,
we revert commit d53c03b0989f8354a7e4dbb947a150fc7fe3f6d1 and call
print_general_info() when no device is specified.

Signed-off-by: Brian Norris <computersforpeace@gmail.com>
---
 ubi-utils/mtdinfo.c |   14 ++++++++++----
 1 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/ubi-utils/mtdinfo.c b/ubi-utils/mtdinfo.c
index d919673..ead4bce 100644
--- a/ubi-utils/mtdinfo.c
+++ b/ubi-utils/mtdinfo.c
@@ -340,7 +340,8 @@ static int print_dev_info(libmtd_t libmtd, const struct mtd_info *mtd_info, int
 	return 0;
 }
 
-static int print_general_info(libmtd_t libmtd, const struct mtd_info *mtd_info)
+static int print_general_info(libmtd_t libmtd, const struct mtd_info *mtd_info,
+			      int all)
 {
 	int i, err, first = 1;
 	struct mtd_dev_info mtd;
@@ -367,9 +368,14 @@ static int print_general_info(libmtd_t libmtd, const struct mtd_info *mtd_info)
 		}
 	}
 	printf("\n");
-	printf("Sysfs interface supported:      %s\n\n",
+	printf("Sysfs interface supported:      %s\n",
 	       mtd_info->sysfs_supported ? "yes" : "no");
 
+	if (!all)
+		return 0;
+
+	printf("\n");
+
 	for (i = mtd_info->lowest_mtd_num;
 	     i <= mtd_info->highest_mtd_num; i++) {
 		err = print_dev_info(libmtd, mtd_info, i);
@@ -404,7 +410,7 @@ int main(int argc, char * const argv[])
 		return sys_errmsg("cannot get MTD information");
 	}
 
-	if (!args.all) {
+	if (!args.all && args.node) {
 		int mtdn;
 
 		/*
@@ -416,7 +422,7 @@ int main(int argc, char * const argv[])
 			goto out_libmtd;
 		err = print_dev_info(libmtd, &mtd_info, mtdn);
 	} else
-		err = print_general_info(libmtd, &mtd_info);
+		err = print_general_info(libmtd, &mtd_info, args.all);
 	if (err)
 		goto out_libmtd;
 
-- 
1.7.5.4

Re: [PATCH 1/2] mtd-utils: add jffs2reader to .gitignore

[Mike Frysinger]

Acked-by: Mike Frysinger <vapier@gentoo.org>
-mike

Re: [PATCH 2/2] mtdinfo: provide info when used without arguments

[Mike Frysinger]

Acked-by: Mike Frysinger <vapier@gentoo.org>
-mike

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

On Wed, 2011-11-16 at 12:03 -0800, Brian Norris wrote:
> Speaking of releases: when will we make a new release? After kernel 3.2
> is final? Or perhaps a pre-3.2 and post-3.2 release? There are a lot of
> important fixes in git right now, and I'd like to incorporate them
> officially.

We can release it as soon as someone ask. Should I do this now?

> Also, note that the following commit was intended to be paired with the
> commits in the "brian" branch:
> 
>     commit bf01f2960ba82468b1b25f00e044fd0c3ee0770a
>     nandwrite: consolidate buffer usage
> 
> While it is functionally harmless, it does include an unnecessary
> memcpy(). In fact, the patch really is useful only when paired with the
> rest of the 'brian' branch, which is intended for kernels supporting
> ioctl(MEMWRITE) - i.e., kernel version >= 3.2.
> 
> So I recommend that the next release either includes the 'brian' branch
> or does not include commit bf01f2. Mike Frysinger and I discussed this
> issue on the ML earlier and didn't really come up with a solution that
> avoids the overhead both with and without MEMWRITE. I figure pre-3.2
> kernel users will just have to put up with it eventually.

Could you please refresh my memory and remind me about the brian branch?
Why we have it and what prevents us from merging it to the master
branch? There were some kernel changes which I was not sure about? But
now they are in Linus' tree so we can merge that branch and make a
release, right?

Artem.

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

On Wed, 2011-11-16 at 12:03 -0800, Brian Norris wrote:
> Hi,
> 
> These two fixes are somewhat trivial, although the second one fixes a
> kind of regression that occurred since the last release.

Pushed both, thanks!

Artem.

Re: [PATCH 0/2] small mtd-utils fixes

[Brian Norris]

On Thu, Nov 17, 2011 at 2:19 PM, Artem Bityutskiy <dedekind1@gmail.com> wrote:
> We can release it as soon as someone ask. Should I do this now?

I suppose I'm asking, so yes :)

> Could you please refresh my memory and remind me about the brian branch?
> Why we have it and what prevents us from merging it to the master
> branch? There were some kernel changes which I was not sure about? But
> now they are in Linus' tree so we can merge that branch and make a
> release, right?

Your exact quote: "And last but not least, I think we should not push
your changes to the master branch of mtd-utils _befor_ the
corresponding kernel-space changes have reached Linuses tree (e.g.,
dwmw2 might dislike something)."

So I suppose it was just waiting until the kernel changes passed
dwmw2. Now's a good time to merge 'brian' and release, I guess.

Thanks,
Brian

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

On Thu, 2011-11-17 at 15:56 -0800, Brian Norris wrote:
> On Thu, Nov 17, 2011 at 2:19 PM, Artem Bityutskiy <dedekind1@gmail.com> wrote:
> > We can release it as soon as someone ask. Should I do this now?
> 
> I suppose I'm asking, so yes :)

Done: mtd-utils-1.4.7
The tarball is at the usual place as well.

Artem.

Re: [PATCH 0/2] small mtd-utils fixes

[Peter Korsgaard]

>>>>> "Artem" == Artem Bityutskiy <dedekind1@gmail.com> writes:

 Artem> On Thu, 2011-11-17 at 15:56 -0800, Brian Norris wrote:
 >> On Thu, Nov 17, 2011 at 2:19 PM, Artem Bityutskiy <dedekind1@gmail.com> wrote:
 >> > We can release it as soon as someone ask. Should I do this now?
 >> 
 >> I suppose I'm asking, so yes :)

 Artem> Done: mtd-utils-1.4.7
 Artem> The tarball is at the usual place as well.

Thanks, I've updated the package in buildroot. One minor nit, you forgot
to update the version in the Makefile.

-- 
Bye, Peter Korsgaard

Re: [PATCH 0/2] small mtd-utils fixes

[Brian Norris]

On Fri, Nov 18, 2011 at 2:44 PM, Peter Korsgaard <jacmet@sunsite.dk> wrote:
> Thanks, I've updated the package in buildroot. One minor nit, you forgot
> to update the version in the Makefile.

I was just about to comment on the Makefile version as well; I forgot
about that...any chance we can fix this on the official release? It's
kind of important that I can tell which version my users are using,
now that we've fixed a lot of annoying bugs/"features". If not, I may
just modify it in my own releases.

Brian

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

On Fri, 2011-11-18 at 15:37 -0800, Brian Norris wrote:
> On Fri, Nov 18, 2011 at 2:44 PM, Peter Korsgaard <jacmet@sunsite.dk> wrote:
> > Thanks, I've updated the package in buildroot. One minor nit, you forgot
> > to update the version in the Makefile.
> 
> I was just about to comment on the Makefile version as well; I forgot
> about that...any chance we can fix this on the official release? It's
> kind of important that I can tell which version my users are using,
> now that we've fixed a lot of annoying bugs/"features". If not, I may
> just modify it in my own releases.

My apologies. I've released 1.4.8 and changed the Makefile this time.
Uploaded the tarball as well.

Artem.

Re: [PATCH 0/2] small mtd-utils fixes

[Mike Frysinger]

On Sat, Nov 19, 2011 at 15:47, Artem Bityutskiy wrote:
> On Fri, 2011-11-18 at 15:37 -0800, Brian Norris wrote:
>> On Fri, Nov 18, 2011 at 2:44 PM, Peter Korsgaard wrote:
>> > Thanks, I've updated the package in buildroot. One minor nit, you forgot
>> > to update the version in the Makefile.
>>
>> I was just about to comment on the Makefile version as well; I forgot
>> about that...any chance we can fix this on the official release? It's
>> kind of important that I can tell which version my users are using,
>> now that we've fixed a lot of annoying bugs/"features". If not, I may
>> just modify it in my own releases.
>
> My apologies. I've released 1.4.8 and changed the Makefile this time.
> Uploaded the tarball as well.

how are you making the releases ?  seems like this keeps coming up, so
maybe we should have a `make dist` that takes care of this stuff
automatically to avoid future mistakes ...
-mike

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

On Sat, 2011-11-19 at 15:55 -0500, Mike Frysinger wrote:
> On Sat, Nov 19, 2011 at 15:47, Artem Bityutskiy wrote:
> > On Fri, 2011-11-18 at 15:37 -0800, Brian Norris wrote:
> >> On Fri, Nov 18, 2011 at 2:44 PM, Peter Korsgaard wrote:
> >> > Thanks, I've updated the package in buildroot. One minor nit, you forgot
> >> > to update the version in the Makefile.
> >>
> >> I was just about to comment on the Makefile version as well; I forgot
> >> about that...any chance we can fix this on the official release? It's
> >> kind of important that I can tell which version my users are using,
> >> now that we've fixed a lot of annoying bugs/"features". If not, I may
> >> just modify it in my own releases.
> >
> > My apologies. I've released 1.4.8 and changed the Makefile this time.
> > Uploaded the tarball as well.
> 
> how are you making the releases ?  seems like this keeps coming up, so
> maybe we should have a `make dist` that takes care of this stuff
> automatically to avoid future mistakes ...

You are right. I've just created a simple shell script which should do
cut a new release properly.

#!/bin/sh -uef

# A small helper script to release mtd-utils. Takes the new version
# as a parameter.

fatal() {
	printf "%s\n" "$1" >&2
	exit 1
}

[ $# -eq 1 ] || fatal "Please, specify the new version"
new_ver="$1"

# Change the version in the Makefile
tmpfile="$(mktemp)"
sed -e "s/^VERSION = [[:digit:]]\+\.[[:digit:]]\+\.[[:digit:]]\+/VERSION = $new_ver/" \
	Makefile > "$tmpfile"
mv "$tmpfile" Makefile

# And commit the change
git commit -s -m "Release mtd-utils-$new_ver" Makefile

# Create new tag
git tag "v$new_ver"

# Prepare the tarball
git archive --format=tar --prefix="mtd-utils-$new_ver/" "v$new_ver" | \
	bzip2 > "mtd-utils-$new_ver.tar.bz2"

# Upload the tarball
scp "mtd-utils-$new_ver.tar.bz2" casper.infradead.org:/var/ftp/pub/mtd-utils/

Re: [PATCH 0/2] small mtd-utils fixes

[Mike Frysinger]

On Sat, Nov 19, 2011 at 18:04, Artem Bityutskiy wrote:
> On Sat, 2011-11-19 at 15:55 -0500, Mike Frysinger wrote:
>> how are you making the releases ?  seems like this keeps coming up, so
>> maybe we should have a `make dist` that takes care of this stuff
>> automatically to avoid future mistakes ...
>
> You are right. I've just created a simple shell script which should do
> cut a new release properly.

could you add it to the git repo ?  be easier to send fixes/tweaks :).
-mike

Re: [PATCH 0/2] small mtd-utils fixes

[David Woodhouse]

[-- Attachment #1: Type: text/plain, Size: 419 bytes --]

On Sun, 2011-11-20 at 01:04 +0200, Artem Bityutskiy wrote:
> # Prepare the tarball
> git archive --format=tar --prefix="mtd-utils-$new_ver/" "v$new_ver" | \
>         bzip2 > "mtd-utils-$new_ver.tar.bz2"

gpg --detach-sign -a "mtd-utils-$new_ver.tar.bz2"

> # Upload the tarball
> scp "mtd-utils-$new_ver.tar.bz2" casper.infradead.org:/var/ftp/pub/mtd-utils/

mtd-utils-$new_ver.tar.bz2{.asc,}

-- 
dwmw2

[-- Attachment #2: smime.p7s --]
[-- Type: application/x-pkcs7-signature, Size: 5818 bytes --]

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

On Sun, 2011-11-20 at 11:55 +0000, David Woodhouse wrote:
> On Sun, 2011-11-20 at 01:04 +0200, Artem Bityutskiy wrote:
> > # Prepare the tarball
> > git archive --format=tar --prefix="mtd-utils-$new_ver/" "v$new_ver" | \
> >         bzip2 > "mtd-utils-$new_ver.tar.bz2"
> 
> gpg --detach-sign -a "mtd-utils-$new_ver.tar.bz2"
> 
> > # Upload the tarball
> > scp "mtd-utils-$new_ver.tar.bz2" casper.infradead.org:/var/ftp/pub/mtd-utils/
> 
> mtd-utils-$new_ver.tar.bz2{.asc,}

OK, I've updated the script:

1. Added some sanity checks
2. Sign tags
3. Sign the tarball
4. Instead of uploading, print the instructions how to do this

#!/bin/sh -uef

# A small helper script to release mtd-utils. Takes the new version
# as a parameter.

fatal() {
	printf "Error: %s\n" "$1" >&2
	exit 1
}

usage() {
	cat <<EOF
Usage: ${0##*/} <new_ver> <outdir>

<new_ver>  - mtd utils version to create in X.Y.Z format
<outdir>   - the output directory where to store the tarball with the
             gpg signature
EOF
	exit 0
}

[ $# -eq 0 ] && usage
[ $# -eq 2 ] || fatal "Insufficient or too many argumetns"

new_ver="$1"; shift
outdir="$1"; shift

release_name="mtd-utils-$new_ver"
tag_name="v$new_ver"

# Make sure the input is sane and the makefile contains sensible version
echo "$new_ver" | egrep -q -x '[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+' ||
	fatal "please, provide new version in X.Y.Z format"

egrep -q -x 'VERSION = [[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+' Makefile ||
	fatal "Makefile does not contain \"Version = X.Y.Z\" line"

# Make sure the git index is up-to-date
[ -z "$(git status --porcelain)" ] || fatal "Git index is not up-to-date"

# Make sure the tag does not exist
[ -z "$(git tag -l "$tag_name")" ] || fatal "Tag $tag_name already exists"

# Change the version in the Makefile
tmpfile="$(mktemp)"
sed -e "s/^VERSION = [[:digit:]]\+\.[[:digit:]]\+\.[[:digit:]]\+/VERSION = $new_ver/" \
	Makefile > "$tmpfile"
mv "$tmpfile" Makefile

# And commit the change
git commit -s -m "Release $release_name" Makefile

# Create new signed tag
git tag -m "$release_name" -s "$tag_name"

# Prepare signed tarball
git archive --format=tar --prefix="$release_name/" "$tag_name" | \
	bzip2 > "$outdir/$release_name.tar.bz2"
gpg --detach-sign -a "$outdir/$release_name.tar.bz2" > "$outdir/$release_name.tar.bz2.asc"

echo "Created $outdir/$release_name.tar.bz2"
echo "Please, verify, then push the tag and upload the tarball and the signature"
echo "You can use these commands:"
echo "git push $tag_name"
echo "scp $outdir/$release_name.tar.bz2 $outdir/$release_name.tar.bz2.asc casper.infradead.org:/var/ftp/pub/mtd-utils"

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

On Sun, 2011-11-20 at 03:41 -0500, Mike Frysinger wrote:
> On Sat, Nov 19, 2011 at 18:04, Artem Bityutskiy wrote:
> > On Sat, 2011-11-19 at 15:55 -0500, Mike Frysinger wrote:
> >> how are you making the releases ?  seems like this keeps coming up, so
> >> maybe we should have a `make dist` that takes care of this stuff
> >> automatically to avoid future mistakes ...
> >
> > You are right. I've just created a simple shell script which should do
> > cut a new release properly.
> 
> could you add it to the git repo ?  be easier to send fixes/tweaks :).

Yeah, I'll do this when everyone is OK with the initial version.

Artem.

Re: [PATCH 0/2] small mtd-utils fixes

[Mike Frysinger]

On Sun, Nov 20, 2011 at 08:08, Artem Bityutskiy wrote:
> usage() {
>        cat <<EOF
> Usage: ${0##*/} <new_ver> <outdir>
>
> <new_ver>  - mtd utils version to create in X.Y.Z format
> <outdir>   - the output directory where to store the tarball with the
>             gpg signature
> EOF
>        exit 0
> }

i'd have outdir be optional and default to $PWD

prob want to update .gitignore to ignore "mtd-utils-*.tar*"

> # Change the version in the Makefile
> tmpfile="$(mktemp)"
> sed -e "s/^VERSION = [[:digit:]]\+\.[[:digit:]]\+\.[[:digit:]]\+/VERSION = $new_ver/" \
>        Makefile > "$tmpfile"
> mv "$tmpfile" Makefile

use `sed -i` and you won't need the $tmpfile at all

> gpg --detach-sign -a "$outdir/$release_name.tar.bz2" > "$outdir/$release_name.tar.bz2.asc"

by using --detach-sign, you don't want the output redirection.  so
delete the "> .......".

> echo "Created $outdir/$release_name.tar.bz2"
> echo "Please, verify, then push the tag and upload the tarball and the signature"
> echo "You can use these commands:"
> echo "git push $tag_name"
> echo "scp $outdir/$release_name.tar.bz2 $outdir/$release_name.tar.bz2.asc casper.infradead.org:/var/ftp/pub/mtd-utils"

if you really like, you could make this a single cat like in the usage()
-mike

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

On Sun, 2011-11-20 at 23:22 -0500, Mike Frysinger wrote:
> On Sun, Nov 20, 2011 at 08:08, Artem Bityutskiy wrote:
> > usage() {
> >        cat <<EOF
> > Usage: ${0##*/} <new_ver> <outdir>
> >
> > <new_ver>  - mtd utils version to create in X.Y.Z format
> > <outdir>   - the output directory where to store the tarball with the
> >             gpg signature
> > EOF
> >        exit 0
> > }
> 
> i'd have outdir be optional and default to $PWD
> 
> prob want to update .gitignore to ignore "mtd-utils-*.tar*"

Yeah, but this would require me to add options handling using getopt,
because optional parameters have to have an option. I just do not feel
it is necessary right now. This can be done at any point if someone
needs.

> 
> > # Change the version in the Makefile
> > tmpfile="$(mktemp)"
> > sed -e "s/^VERSION = [[:digit:]]\+\.[[:digit:]]\+\.[[:digit:]]\+/VERSION = $new_ver/" \
> >        Makefile > "$tmpfile"
> > mv "$tmpfile" Makefile
> 
> use `sed -i` and you won't need the $tmpfile at all

Good point, thanks, I'll do this.

> 
> > gpg --detach-sign -a "$outdir/$release_name.tar.bz2" > "$outdir/$release_name.tar.bz2.asc"
> 
> by using --detach-sign, you don't want the output redirection.  so
> delete the "> .......".

OK, I'll actually test this and fix.

> > echo "Created $outdir/$release_name.tar.bz2"
> > echo "Please, verify, then push the tag and upload the tarball and the signature"
> > echo "You can use these commands:"
> > echo "git push $tag_name"
> > echo "scp $outdir/$release_name.tar.bz2 $outdir/$release_name.tar.bz2.asc casper.infradead.org:/var/ftp/pub/mtd-utils"
> 
> if you really like, you could make this a single cat like in the usage()

Will do.

But now I have to go sleep, will send an updated patch in near future.
Thank you for suggestions.

Artem.

Re: [PATCH 0/2] small mtd-utils fixes

[Mike Frysinger]

On Tue, Nov 22, 2011 at 16:38, Artem Bityutskiy wrote:
> On Sun, 2011-11-20 at 23:22 -0500, Mike Frysinger wrote:
>> On Sun, Nov 20, 2011 at 08:08, Artem Bityutskiy wrote:
>> > usage() {
>> >        cat <<EOF
>> > Usage: ${0##*/} <new_ver> <outdir>
>> >
>> > <new_ver>  - mtd utils version to create in X.Y.Z format
>> > <outdir>   - the output directory where to store the tarball with the
>> >             gpg signature
>> > EOF
>> >        exit 0
>> > }
>>
>> i'd have outdir be optional and default to $PWD
>
> Yeah, but this would require me to add options handling using getopt,
> because optional parameters have to have an option. I just do not feel
> it is necessary right now. This can be done at any point if someone
> needs.

i don't think so

case $# in
  0) usage ;;
  1|2) ;;
  *) fatal "Insufficient or too many argumetns" ;;
esac

new_ver="$1"; shift
outdir="${1:-.}"; shift
-mike

Re: [PATCH 0/2] small mtd-utils fixes

[Mike Frysinger]

On Sun, Nov 20, 2011 at 08:08, Artem Bityutskiy wrote:
> [ $# -eq 2 ] || fatal "Insufficient or too many argumetns"

typo: arguments
-mike

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

[-- Attachment #1: Type: text/plain, Size: 1420 bytes --]

On Tue, 2011-11-22 at 17:21 -0500, Mike Frysinger wrote:
> On Tue, Nov 22, 2011 at 16:38, Artem Bityutskiy wrote:
> > On Sun, 2011-11-20 at 23:22 -0500, Mike Frysinger wrote:
> >> On Sun, Nov 20, 2011 at 08:08, Artem Bityutskiy wrote:
> >> > usage() {
> >> >        cat <<EOF
> >> > Usage: ${0##*/} <new_ver> <outdir>
> >> >
> >> > <new_ver>  - mtd utils version to create in X.Y.Z format
> >> > <outdir>   - the output directory where to store the tarball with the
> >> >             gpg signature
> >> > EOF
> >> >        exit 0
> >> > }
> >>
> >> i'd have outdir be optional and default to $PWD
> >
> > Yeah, but this would require me to add options handling using getopt,
> > because optional parameters have to have an option. I just do not feel
> > it is necessary right now. This can be done at any point if someone
> > needs.
> 
> i don't think so
> 
> case $# in
>   0) usage ;;
>   1|2) ;;
>   *) fatal "Insufficient or too many argumetns" ;;
> esac
> 
> new_ver="$1"; shift
> outdir="${1:-.}"; shift

I did not mean it is technincally impossible, I meant that it is just
normal Unix rule which I want to stick with, at least in simple cases:

1. Non-mandatory arguments come with options like -x <blah>
2. Mandatory arguments come come without options.

utility < -x blah and other optional stuff> <arg1> <other mandatory
args>

-- 
Best Regards,
Artem Bityutskiy

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

[-- Attachment #1: Type: text/plain, Size: 2205 bytes --]

Hi, updated the script as you suggested. Did not do the "outdir should
be optional" part because it is not needed now - we can do this later if
we want.

#!/bin/sh -uef

# A small helper script to release mtd-utils. Takes the new version
# as a parameter.

fatal() {
	printf "Error: %s\n" "$1" >&2
	exit 1
}

usage() {
	cat <<EOF
Usage: ${0##*/} <new_ver> <outdir>

<new_ver>  - mtd utils version to create in X.Y.Z format
<outdir>   - the output directory where to store the tarball with the
             gpg signature
EOF
	exit 0
}

[ $# -eq 0 ] && usage
[ $# -eq 2 ] || fatal "Insufficient or too many argumetns"

new_ver="$1"; shift
outdir="$1"; shift

release_name="mtd-utils-$new_ver"
tag_name="v$new_ver"

# Make sure the input is sane and the makefile contains sensible version
echo "$new_ver" | egrep -q -x '[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+' ||
	fatal "please, provide new version in X.Y.Z format"

egrep -q -x 'VERSION = [[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+' Makefile ||
	fatal "Makefile does not contain \"Version = X.Y.Z\" line"

# Make sure the git index is up-to-date
[ -z "$(git status --porcelain)" ] || fatal "Git index is not up-to-date"

# Make sure the tag does not exist
[ -z "$(git tag -l "$tag_name")" ] || fatal "Tag $tag_name already exists"

# Change the version in the Makefile
sed -i -e "s/^VERSION = [[:digit:]]\+\.[[:digit:]]\+\.[[:digit:]]\+/VERSION = $new_ver/" Makefile

# And commit the change
git commit -s -m "Release $release_name" Makefile

# Create new signed tag
echo "Signing tag $tag_name"
git tag -m "$release_name" -s "$tag_name"

# Prepare signed tarball
git archive --format=tar --prefix="$release_name/" "$tag_name" | \
	bzip2 > "$outdir/$release_name.tar.bz2"
echo "Signing the tarball"
gpg -o "$outdir/$release_name.tar.bz2.asc" --detach-sign -a "$outdir/$release_name.tar.bz2"

cat <<EOF
Created $outdir/$release_name.tar.bz2
Please, verify, then push the tag and upload the tarball and the signature
You can use these commands:
git push origin $tag_name
scp $outdir/$release_name.tar.bz2 $outdir/$release_name.tar.bz2.asc casper.infradead.org:/var/ftp/pub/mtd-utils
EOF


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

[-- Attachment #1: Type: text/plain, Size: 270 bytes --]

On Tue, 2011-11-22 at 17:21 -0500, Mike Frysinger wrote:
> On Sun, Nov 20, 2011 at 08:08, Artem Bityutskiy wrote:
> > [ $# -eq 2 ] || fatal "Insufficient or too many argumetns"
> 
> typo: arguments

Oh, yes, will also fix.

-- 
Best Regards,
Artem Bityutskiy

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 0/2] small mtd-utils fixes

[Mike Frysinger]

On Wed, Nov 23, 2011 at 02:45, Artem Bityutskiy wrote:
> On Tue, 2011-11-22 at 17:21 -0500, Mike Frysinger wrote:
>> On Tue, Nov 22, 2011 at 16:38, Artem Bityutskiy wrote:
>> > On Sun, 2011-11-20 at 23:22 -0500, Mike Frysinger wrote:
>> >> On Sun, Nov 20, 2011 at 08:08, Artem Bityutskiy wrote:
>> >> > usage() {
>> >> >        cat <<EOF
>> >> > Usage: ${0##*/} <new_ver> <outdir>
>> >> >
>> >> > <new_ver>  - mtd utils version to create in X.Y.Z format
>> >> > <outdir>   - the output directory where to store the tarball with the
>> >> >             gpg signature
>> >> > EOF
>> >> >        exit 0
>> >> > }
>> >>
>> >> i'd have outdir be optional and default to $PWD
>> >
>> > Yeah, but this would require me to add options handling using getopt,
>> > because optional parameters have to have an option. I just do not feel
>> > it is necessary right now. This can be done at any point if someone
>> > needs.
>>
>> i don't think so
>>
>> case $# in
>>   0) usage ;;
>>   1|2) ;;
>>   *) fatal "Insufficient or too many argumetns" ;;
>> esac
>>
>> new_ver="$1"; shift
>> outdir="${1:-.}"; shift
>
> I did not mean it is technincally impossible, I meant that it is just
> normal Unix rule which I want to stick with, at least in simple cases:
>
> 1. Non-mandatory arguments come with options like -x <blah>
> 2. Mandatory arguments come come without options.
>
> utility < -x blah and other optional stuff> <arg1> <other mandatory
> args>

sure, but we're not taking about options here, we're talking about
required arguments:
./make-release.sh <version> [output_dir]

if we were to make it into:
./make-release.sh [-o output_dir] <version>

then i'd understand the `getopt` desire
-mike

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

[-- Attachment #1: Type: text/plain, Size: 597 bytes --]

On Wed, 2011-11-23 at 20:45 -0500, Mike Frysinger wrote:
> sure, but we're not taking about options here, we're talking about
> required arguments:
> ./make-release.sh <version> [output_dir]

Yes, but my point was that this way is not very normal and not
"standard" for Unix, at least by my perception of "standard".

> if we were to make it into:
> ./make-release.sh [-o output_dir] <version>
> 
> then i'd understand the `getopt` desire

This is something standard - options are optional and go first, then
mandatory stuff without options.

-- 
Best Regards,
Artem Bityutskiy

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 0/2] small mtd-utils fixes

[Mike Frysinger]

On Thu, Nov 24, 2011 at 02:26, Artem Bityutskiy wrote:
> On Wed, 2011-11-23 at 20:45 -0500, Mike Frysinger wrote:
>> sure, but we're not taking about options here, we're talking about
>> required arguments:
>> ./make-release.sh <version> [output_dir]
>
> Yes, but my point was that this way is not very normal and not
> "standard" for Unix, at least by my perception of "standard".

for tools installed into $PATH and expected a little "higher" terms of
quality, then sure, it's not standard.  for tiny scripts in repo dirs
that serve specific functions, i think it's fairly common.

i think it should be the way i suggested, but i won't fight over it
too much ... i'm not the one making releases after all.
-mike

Re: [PATCH 0/2] small mtd-utils fixes

[Artem Bityutskiy]

[-- Attachment #1: Type: text/plain, Size: 410 bytes --]

On Sun, 2011-12-04 at 19:28 -0500, Mike Frysinger wrote:
> i think it should be the way i suggested, but i won't fight over it
> too much ... i'm not the one making releases after all.

I've added the script to the mtd-utils.git repo. Thanks for review. I
nevertheless did not change that parameter thing - we can do it when /
if we actually need this.

Artem.

-- 
Best Regards,
Artem Bityutskiy

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]