[arch-projects] netcfg: fixes and features
Dear Thomas and the rest, A few of us have been looking at the netcfg code lately. One thing that came to light was that FS#24599 is actualy a wpa_actiond bug, so, Thomas, could you take it upon you? Furthermore, Remy's development tree held 8 bugfixes and some improvements, which I augmented with another 5.5 bugfixes and 3.5 added features (the half one being a syntax to reconnect based on an interface, as a fix for FS#28196). I would like to see someone package this work, that is available at https://github.com/joukewitteveen/netcfg and push a package to the (testing?) repository. I am not requesting maintainership of the code (Alfredo Palhares is after it too), but I am willing to be the maintainer. My request is just to move netcfg forward. Regards, - Jouke List of fixed bugs outside Remy's tree: FS#23934 FS#26370 FS#26607 FS#27496 (not quite) FS#28022 FS#28196
Am 28.02.2012 00:31, schrieb Jouke Witteveen:
Dear Thomas and the rest,
A few of us have been looking at the netcfg code lately. One thing that came to light was that FS#24599 is actualy a wpa_actiond bug, so, Thomas, could you take it upon you?
I looked into it shortly, the last comment is correct.
Furthermore, Remy's development tree held 8 bugfixes and some improvements, which I augmented with another 5.5 bugfixes and 3.5 added features (the half one being a syntax to reconnect based on an interface, as a fix for FS#28196).
I would like to see someone package this work, that is available at https://github.com/joukewitteveen/netcfg and push a package to the (testing?) repository.
I am not requesting maintainership of the code (Alfredo Palhares is after it too), but I am willing to be the maintainer. My request is just to move netcfg forward.
Both of you, I'd like to point your attention towards https://mailman.archlinux.org/pipermail/arch-projects/2011-December/002254.h... if you haven't merged it already (the patch is also on my github). Now, we need to make one of you the maintainer of the official netcfg tree on al.org. I don't know which one of you, so you two fight it out and tell us the result. Or, anyone else on this list, choose. I don't care, as long as I don't have to decide.
On Tue, Feb 28, 2012 at 08:59, Thomas Bächler <thomas@archlinux.org> wrote:
Am 28.02.2012 00:31, schrieb Jouke Witteveen:
Dear Thomas and the rest,
A few of us have been looking at the netcfg code lately. One thing that came to light was that FS#24599 is actualy a wpa_actiond bug, so, Thomas, could you take it upon you?
I looked into it shortly, the last comment is correct.
That comment was mine. I wondered whether there was a reason to look at occurrences of ']'.
Furthermore, Remy's development tree held 8 bugfixes and some improvements, which I augmented with another 5.5 bugfixes and 3.5 added features (the half one being a syntax to reconnect based on an interface, as a fix for FS#28196).
I would like to see someone package this work, that is available at https://github.com/joukewitteveen/netcfg and push a package to the (testing?) repository.
I am not requesting maintainership of the code (Alfredo Palhares is after it too), but I am willing to be the maintainer. My request is just to move netcfg forward.
Both of you, I'd like to point your attention towards https://mailman.archlinux.org/pipermail/arch-projects/2011-December/002254.h... if you haven't merged it already (the patch is also on my github).
That one was merged in 7ba4 last week :-).
Now, we need to make one of you the maintainer of the official netcfg tree on al.org. I don't know which one of you, so you two fight it out and tell us the result. Or, anyone else on this list, choose. I don't care, as long as I don't have to decide.
Will do. Thanks!
Am 28.02.2012 09:57, schrieb Jouke Witteveen:
On Tue, Feb 28, 2012 at 08:59, Thomas Bächler <thomas@archlinux.org> wrote:
Am 28.02.2012 00:31, schrieb Jouke Witteveen:
Dear Thomas and the rest,
A few of us have been looking at the netcfg code lately. One thing that came to light was that FS#24599 is actualy a wpa_actiond bug, so, Thomas, could you take it upon you?
I looked into it shortly, the last comment is correct.
That comment was mine. I wondered whether there was a reason to look at occurrences of ']'.
The lack of a proper API to the wpa_supplicant management interface forces us to parse the output of the status notifications ourselves. The notification in question looks like this: <2>CTRL-EVENT-CONNECTED - Connection to 00:11:22:33:44:55 completed (reauth) [id=10 id_str=profile_name] This format is about as primitive and broken as my parser. Given that this is only a single line, the patch I'll send in a minute SHOULD solve this, but I can't test this anytime soon.
Fixes FS#24599 --- wpa_actiond.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/wpa_actiond.c b/wpa_actiond.c index 84eef63..69eb6f2 100644 --- a/wpa_actiond.c +++ b/wpa_actiond.c @@ -188,7 +188,7 @@ static void parse_wpa_event(const char *msg, enum wpaevent *ev, char *idstr) pos = strstr(pos, "id_str="); if(pos != NULL) { pos += strlen("id_str="); - pos2 = index(pos, ']'); + pos2 = rindex(pos, ']'); strncpy(idstr, pos, min(pos2-pos, IDSTRLEN-1)); idstr[min(pos2-pos, IDSTRLEN-1)] = '\0'; } -- 1.7.9.2
Dear Thomas and the rest,
I would like to see someone package this work, that is available at https://github.com/joukewitteveen/netcfg and push a package to the (testing?) repository.
I am not requesting maintainership of the code (Alfredo Palhares is after it too), but I am willing to be the maintainer. My request is just to move netcfg forward.
Now, we need to make one of you the maintainer of the official netcfg tree on al.org. I don't know which one of you, so you two fight it out and tell us the result. Or, anyone else on this list, choose. I don't care, as long as I don't have to decide.
We have decided that I will take on maintainership and Alfredo and I will keep in close contact for further development. The tree at https://github.com/joukewitteveen/netcfg has seen some fairly big changes lately, additionally fixing FS#25587 and FS#27496. The new changes require a new PKGBUILD, which is included below (but likely broken by Gmail). This is just for completeness to anyone curious. I hope to get some details on maintainership soon. Regards, - Jouke --- # $Id$ # Maintainer: Jouke Witteveen <j.witteveen@gmail.com> pkgname=netcfg pkgver=2.7 pkgrel=1 pkgdesc="Network configuration and profile scripts" url="http://archlinux.org" license=("BSD") backup=(etc/iftab etc/conf.d/netcfg) depends=("coreutils" "dhcpcd>=4.0" "iproute2") makedepends=('asciidoc') optdepends=('dialog: Required for menu based profile and wifi selectors' 'bridge-utils: To set up bridge connections' 'wpa_supplicant: required for wireless network support' 'wpa_actiond: Required for automatic wireless connection with rc.d/net-auto-wireless' 'ifplugd: Required for automatic wired connection with rc.d/net-auto-wired' 'wireless_tools: Required for net-rename' ) source=(ftp://ftp.archlinux.org/other/netcfg/netcfg-${pkgver}.tar.gz) arch=(any) md5sums=('ba52544f576f60627ba958e748f1765a') package() { cd "$srcdir/netcfg-${pkgver}" make DESTDIR="$pkgdir" install install -D -m644 LICENSE "$pkgdir/usr/share/licenses/netcfg/LICENSE" # Shell Completion install -D -m644 contrib/bash-completion "$pkgdir/etc/bash_completion.d/netcfg" install -D -m644 contrib/zsh-completion "$pkgdir/usr/share/zsh/site-functions/_netcfg" }
Am 04.03.2012 01:19, schrieb Jouke Witteveen:
Dear Thomas and the rest,
I would like to see someone package this work, that is available at https://github.com/joukewitteveen/netcfg and push a package to the (testing?) repository.
I am not requesting maintainership of the code (Alfredo Palhares is after it too), but I am willing to be the maintainer. My request is just to move netcfg forward.
Now, we need to make one of you the maintainer of the official netcfg tree on al.org. I don't know which one of you, so you two fight it out and tell us the result. Or, anyone else on this list, choose. I don't care, as long as I don't have to decide.
We have decided that I will take on maintainership and Alfredo and I will keep in close contact for further development. The tree at https://github.com/joukewitteveen/netcfg has seen some fairly big changes lately, additionally fixing FS#25587 and FS#27496. The new changes require a new PKGBUILD, which is included below (but likely broken by Gmail). This is just for completeness to anyone curious.
I hope to get some details on maintainership soon.
Regards, - Jouke
So it seems we finally have someone who is willing to work on netcfg. Anyone who is interested in netcfg, please review the git tree. I propose to manually pull the tree and package it on Jouke's pull request. If the results are good, we might talk about direct git-shell access. Thoughts?
Excerpts from Thomas Bächler's message of Sun Mar 04 19:46:40 +0100 2012:
I propose to manually pull the tree and package it on Jouke's pull request. If the results are good, we might talk about direct git-shell access.
Thoughts? Agreed.
Dear Thomas,
So it seems we finally have someone who is willing to work on netcfg. Anyone who is interested in netcfg, please review the git tree.
There has been a slight increase in interest in the GitHub repository, but nothing too spectacular. For my own use-cases, I did not experience any bugs.
I propose to manually pull the tree and package it on Jouke's pull request. If the results are good, we might talk about direct git-shell access.
It has been two weeks since your mail and it is again rather quiet around the GitHub repository. Perhaps the best idea now is to package it for testing and see whether anyone experiences any problems (I don't expect so). Pulling would be nice since many documentation refers to the Arch Linux projects git. Packaging boils down to `make pkgbuild` and `make upload`. The PKGBUILD ends up in the repository root (don't mistakenly try to use the template in contrib/). Regards, - Jouke
Am 18.03.2012 09:49, schrieb Jouke Witteveen:
I propose to manually pull the tree and package it on Jouke's pull request. If the results are good, we might talk about direct git-shell access.
It has been two weeks since your mail and it is again rather quiet around the GitHub repository. Perhaps the best idea now is to package it for testing and see whether anyone experiences any problems (I don't expect so). Pulling would be nice since many documentation refers to the Arch Linux projects git. Packaging boils down to `make pkgbuild` and `make upload`. The PKGBUILD ends up in the repository root (don't mistakenly try to use the template in contrib/).
There has been no reply on arch-dev-public, so I have gone ahead. Your code is up on projects.archlinux.org and a package is in testing. It seems I noticed the 'make pkgbuild' target too late, so please check the ABS PKGBUILD and see if it is okay.
It seems I noticed the 'make pkgbuild' target too late, so please check the ABS PKGBUILD and see if it is okay.
Unfortunately, it is not. One optional dependency (net-tools) was dropped, one build dependency (asciidoc) was added, the optional dependencies dialog and wireless_tools their descriptions changed. Furthermore, the PKGBUILD now deals with installing shell completion. I noticed you changed to sha256sums. Should I upgrade? Regards, - Jouke
On Sun, 18 Mar 2012 21:51:59 +0100 Jouke Witteveen <j.witteveen@gmail.com> wrote:
It seems I noticed the 'make pkgbuild' target too late, so please check the ABS PKGBUILD and see if it is okay.
Unfortunately, it is not. One optional dependency (net-tools) was dropped, one build dependency (asciidoc) was added, the optional dependencies dialog and wireless_tools their descriptions changed. Furthermore, the PKGBUILD now deals with installing shell completion.
I noticed you changed to sha256sums. Should I upgrade?
Regards, - Jouke
Is it OK at all to have a pkg in [core] makedepend on a pkg from [community]? -- Leonid Isaev GnuPG key ID: 164B5A6D Key fingerprint: C0DF 20D0 C075 C3F1 E1BE 775A A7AE F6CB 164B 5A6D
On Sun, Mar 18, 2012 at 22:19, Leonid Isaev <lisaev@umail.iu.edu> wrote:
On Sun, 18 Mar 2012 21:51:59 +0100 Jouke Witteveen <j.witteveen@gmail.com> wrote:
It seems I noticed the 'make pkgbuild' target too late, so please check the ABS PKGBUILD and see if it is okay.
Unfortunately, it is not. One optional dependency (net-tools) was dropped, one build dependency (asciidoc) was added, the optional dependencies dialog and wireless_tools their descriptions changed. Furthermore, the PKGBUILD now deals with installing shell completion.
I noticed you changed to sha256sums. Should I upgrade?
Regards, - Jouke
Is it OK at all to have a pkg in [core] makedepend on a pkg from [community]?
initscripts has the same makedepend (of course, that doesn't say anything) @Thomas: What to do with all the fixed netcfg bugs? They should be closed when netcfg hits [core]. Also, you could try the fix for wpa_actiond in [testing], as it does fix a bug ;-). - Jouke
Am 18.03.2012 22:25, schrieb Jouke Witteveen:
Unfortunately, it is not. One optional dependency (net-tools) was dropped, one build dependency (asciidoc) was added, the optional dependencies dialog and wireless_tools their descriptions changed. Furthermore, the PKGBUILD now deals with installing shell completion.
I'll update it again tomorrow.
I noticed you changed to sha256sums. Should I upgrade?
I use them almost everywhere, but it shouldn't matter.
Is it OK at all to have a pkg in [core] makedepend on a pkg from [community]?
It happens.
@Thomas: What to do with all the fixed netcfg bugs? They should be closed when netcfg hits [core].
Yes, post me a list of bug numbers when that is done.
Also, you could try the fix for wpa_actiond in [testing], as it does fix a bug ;-).
I forgot about that, I still have the fix in my working branch only.
On Sun, Mar 18, 2012 at 10:25 PM, Jouke Witteveen <j.witteveen@gmail.com> wrote:
initscripts has the same makedepend (of course, that doesn't say anything)
Hm, this is not ideal (but since it is only a makedepend it is not that critical either). I should change it so the manpages are in the tarball from "upstream". -t
On 19/03/12 07:19, Leonid Isaev wrote:
On Sun, 18 Mar 2012 21:51:59 +0100 Jouke Witteveen <j.witteveen@gmail.com> wrote:
It seems I noticed the 'make pkgbuild' target too late, so please check the ABS PKGBUILD and see if it is okay.
Unfortunately, it is not. One optional dependency (net-tools) was dropped, one build dependency (asciidoc) was added, the optional dependencies dialog and wireless_tools their descriptions changed. Furthermore, the PKGBUILD now deals with installing shell completion.
I noticed you changed to sha256sums. Should I upgrade?
Regards, - Jouke
Is it OK at all to have a pkg in [core] makedepend on a pkg from [community]?
Not really... but is that really needed? Most projects pregenerate their documentation and include it in the release tarball. Allan
Le 18 mars 2012 21:51, Jouke Witteveen <j.witteveen@gmail.com> a écrit :
It seems I noticed the 'make pkgbuild' target too late, so please check the ABS PKGBUILD and see if it is okay.
Unfortunately, it is not. One optional dependency (net-tools) was dropped, one build dependency (asciidoc) was added, the optional dependencies dialog and wireless_tools their descriptions changed. Furthermore, the PKGBUILD now deals with installing shell completion.
I noticed you changed to sha256sums. Should I upgrade?
Some of the latest changes are not documented in the NEWS file: I think the changes in build and dependencies could be mentioned there. Also, the usage has changed since the last commit "renamed an option", or at least made the old one deprecated. I think an announce should be made somewhere, or at least, it should appear in some NEWS or release notes. Rémy.
On Mon, Mar 19, 2012 at 00:12, Rémy Oudompheng <remyoudompheng@gmail.com> wrote:
Le 18 mars 2012 21:51, Jouke Witteveen <j.witteveen@gmail.com> a écrit :
It seems I noticed the 'make pkgbuild' target too late, so please check the ABS PKGBUILD and see if it is okay.
Unfortunately, it is not. One optional dependency (net-tools) was dropped, one build dependency (asciidoc) was added, the optional dependencies dialog and wireless_tools their descriptions changed. Furthermore, the PKGBUILD now deals with installing shell completion.
I noticed you changed to sha256sums. Should I upgrade?
Some of the latest changes are not documented in the NEWS file: I think the changes in build and dependencies could be mentioned there.
Also, the usage has changed since the last commit "renamed an option", or at least made the old one deprecated. I think an announce should be made somewhere, or at least, it should appear in some NEWS or release notes.
Rémy.
I have updated the repository. Pre-built documentation is now supplied, dropping asciidoc from the dependencies. The NEWS file includes some more news. I'm in favor of an announcement somewhere. Additionally, the wiki would need some updating when 2.7 is released. - Jouke
Am 19.03.2012 17:49, schrieb Jouke Witteveen:
On Mon, Mar 19, 2012 at 00:12, Rémy Oudompheng <remyoudompheng@gmail.com> wrote:
Le 18 mars 2012 21:51, Jouke Witteveen <j.witteveen@gmail.com> a écrit :
It seems I noticed the 'make pkgbuild' target too late, so please check the ABS PKGBUILD and see if it is okay.
Unfortunately, it is not. One optional dependency (net-tools) was dropped, one build dependency (asciidoc) was added, the optional dependencies dialog and wireless_tools their descriptions changed. Furthermore, the PKGBUILD now deals with installing shell completion.
I noticed you changed to sha256sums. Should I upgrade?
Some of the latest changes are not documented in the NEWS file: I think the changes in build and dependencies could be mentioned there.
Also, the usage has changed since the last commit "renamed an option", or at least made the old one deprecated. I think an announce should be made somewhere, or at least, it should appear in some NEWS or release notes.
Rémy.
I have updated the repository. Pre-built documentation is now supplied, dropping asciidoc from the dependencies. The NEWS file includes some more news.
I'm in favor of an announcement somewhere. Additionally, the wiki would need some updating when 2.7 is released.
Technically, I released 2.7 already, so I'll tag your changes as 2.7.1. Version numbers don't matter anyway.
Am 19.03.2012 18:04, schrieb Thomas Bächler:
I'm in favor of an announcement somewhere. Additionally, the wiki would need some updating when 2.7 is released.
Technically, I released 2.7 already, so I'll tag your changes as 2.7.1. Version numbers don't matter anyway.
I updated the PKGBUILD and tagged the release as 2.7.1, please pull the tags and commits from the upstream repository.
On Mon, Mar 19, 2012 at 20:07, Thomas Bächler <thomas@archlinux.org> wrote:
Am 19.03.2012 18:04, schrieb Thomas Bächler:
I'm in favor of an announcement somewhere. Additionally, the wiki would need some updating when 2.7 is released.
Technically, I released 2.7 already, so I'll tag your changes as 2.7.1. Version numbers don't matter anyway.
I updated the PKGBUILD and tagged the release as 2.7.1, please pull the tags and commits from the upstream repository.
Today, the path for bash-completion scripts changed. Although I do not use bash completion, I did a quick update of the PKGBUILD and during testing I discovered some errors in the completion script. They have now been fixed (for as far as I could make sense of the weird completion practices). Perhaps it is already time for a 2.7.2? We'll be running out of version numbers this way :-P.
participants (7)
-
Alfredo Palhares
-
Allan McRae
-
Jouke Witteveen
-
Leonid Isaev
-
Rémy Oudompheng
-
Thomas Bächler
-
Tom Gundersen