Created attachment 172696 [details]
This patch does the following:
* Adds DEBUG option to simplify troubleshooting
* Adds debug logging of wpa_supplicant control interface
* Installs the manpage
* Adds USE_QT4=corelib
* Makes wpa_gui executable setuid to allow it to easily connect to wpa_supplicant control socket
* Fixes the problem that wpa_gui sometimes picks up .pid file instead of the control interface
wpa_gui needs wpa_supplicant to create the control interface socket, which is done by this patch:
With the 211213 patch, wpa_gui now seamlessly works with wpa_supplicant from the port security/wpa_supplicant when /etc/rc.conf has:
passes poudriere on 10.3
Created attachment 172697 [details]
Added these items:
* Compresses the scalable icon
* Adds INSTALLS_ICONS=yes
@Yuri, thanks for your contribution. As this is a MFH candidate, if you could confirm QA testing across all supported versions and at least one each of i386/amd64 (and testing DEBUG on/off) that would be great
Comment on attachment 172697 [details]
Port is unmaintained, implicit approval
It passes poudriere on 9.3 amd64 and 10.3 amd64. It is very unlikely the patch breaks any other builds because the package version didn't change, and this patch only makes cosmetic changes.
Please note that this is a very important port because this is the only WiFi UI manager on FreeBSD. It is currently broken, this patch with 211213 unbreak it.
Created attachment 172698 [details]
* Added the package message explaining what does wpa_gui expect
(implicit patch approval)
Given the source code changes, the ports importance, and the fact that it requires GUI (heavy dependencies) to test, it would be good to confirm successful build on i386 as well. We cant break quarterlies (any more than they might be already), and the QA bar is higher.
Also, does this issue depend on bug 211213?
The port build doesn't depend on 211213. It only needs 211213 in the runtime.
(In reply to yuri from comment #8)
So to clarify, this issue needs the patch from bug 211213 to land before it can be committed, correct?
No, the port can be committed by itself. wpa_supplicant option from bug 211213 can be set manually in /etc/rc.conf. But it is much easier to use when 211213 is committed.
Thank you Yuri
Pending i386 QA
Passes poudriere on 10.3 amd64, 10.3 i386, 9.3 amd64.
Please commit it.
It expects the running wpa_supplicant from the port
So why dont we add RUN_DEPENDS on security/wpa_supplicant ?
A commit references this bug:
Date: Tue Sep 13 20:18:01 UTC 2016
New revision: 422081
net/wpa_gui: Several improvements
- Adds DEBUG option to simplify troubleshooting
- Adds debug logging of wpa_supplicant control interface
- Installs the manpage
- Adds USE_QT4=corelib
- Makes wpa_gui executable setuid to allow it to easily connect to
wpa_supplicant control socket
- Fixes the problem that wpa_gui sometimes picks up .pid file instead
of the control interface
Submitted by: Yuri Victorovich <email@example.com>
Reviewed by: koobs
Committed. Now, about the MFH: I think it's not necessary. koobs ?
(In reply to Kurt Jaeger from comment #13)
Kurt, thanks for committing this.
You are right, I will add RUN_DEPENDS in a separate bug report.