Bug 217054
| Summary: | dns/void-zones-tools: Update to 1.0.2 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | Ports & Packages | Reporter: | Vidar Karlsen <vidar> | ||||||||
| Component: | Individual Port(s) | Assignee: | Danilo G. Baio <dbaio> | ||||||||
| Status: | Closed FIXED | ||||||||||
| Severity: | Affects Some People | CC: | cyclaero, dbaio, mikael, vidar | ||||||||
| Priority: | --- | Keywords: | patch | ||||||||
| Version: | Latest | Flags: | vidar:
maintainer-feedback+
|
||||||||
| Hardware: | Any | ||||||||||
| OS: | Any | ||||||||||
| Attachments: |
|
||||||||||
It builds fine on armv6. char are unsigned on arm, are you sure you need 'CFLAGS = $(CDEFS) -fsigned-char' on arm? Thanks for testing it on arm. The -fsigned-char comes from upstream. If it works fine without it, we should probably remove it. It mostly works on armv6 :)
There is a bug (copy/paste error) in domainlen for !x86 arch: some files aren't 'parsed' correcly and it results in an empty host/zone local-void.zones.
hosts2zones /tmp/local-void.zones /usr/local/etc/void-zones/x_void_list.txt
Number of Hosts: 0
Number of Zones: 0
--- binutils.h.orig 2017-02-14 19:55:10.775822000 +0100
+++ binutils.h 2017-02-14 19:54:57.719906000 +0100
@@ -365,7 +365,7 @@ typedef long long llong;
return 0;
int l;
- for (l = 0; domain[l] && domain[l] != '|'; l++)
+ for (l = 0; domain[l] && domain[l] != '.'; l++)
;
return l;
}
I submitted a PR:
https://github.com/cyclaero/void-zones-tools/pull/2
(In reply to mikael.urankar from comment #1) Hi, I am the developer of the void-zones-tools. You asked: "... char are unsigned on arm, are you sure you need 'CFLAGS = $(CDEFS) -fsigned-char' on arm?" Strictly speaking, this not necessary. However, my development environment is on a x86 system on which char is signed by default. When I ported my software to arm, I ran into an issue within one quite old function of thousands, where I lazily assumed that char is always signed, as you said, it is not on ARM and not on other architectures. In addition, I got some other projects utilizing the same set of helper functions, which I want to work the same on every supported system. I revisited most of my code and it should be agnostic to the signedness of char. Anyway, I find it important that all the standard C-types are behaving the smae on all the system that I want to support. So, I would have either added -funsigned-char for x86 or -fsigned-char for ARM, the latter of which was much more effortless, and that's why I would like to aak to keep this as it is. Created attachment 180000 [details]
svn diff
New patch to pull the latest commit (bfc0566) from upstream. This obsoletes the old patch.
I built an armv6 jail to test, and it worked fine with latest commit (bfc0566). I threw i386 into the mix too for good measure. I have now successfully run the following tests: portlint -C: ok poudriere testport on 10.3-amd64: ok poudriere testport on 11.0-amd64: ok poudriere testport on 10.3-i386: ok poudriere testport on 11.0-i386: ok poudriere testport on 11.0-arm.armv6: ok I tested that it runs too, in all the test jails. :) Changing the title to reflect the new upstream commit hash, bfc0566. (In reply to cyclaero from comment #4) Hi. Any chance to tag a version in Github? Regards. No, there are no tags on github https://github.com/cyclaero/void-zones-tools (In reply to mikael.urankar from comment #8) Yes, I'm aware. Because of this I asked cyclaero(developer of dns/void-zones-tools) if is it possible to tag a version there. https://www.freebsd.org/doc/en_US.ISO8859-1/books/porters-handbook/makefile-distfiles.html#makefile-master_sites-github-ex4 ^^ If we use this way, will be needed to set `PORTEPOCH=1`. Regards. (In reply to Danilo G. Baio from comment #7) I am ready to tag anything, however, I only know how to tag my SVN repositories. Since I found GIT always doing the things differently than I wanted it to be done, I am trying hard to avoid GIT. Before I destroy my repository in the course of various tagging trials, I would appreciate to be told exactly what needs to be done. (In reply to cyclaero from comment #10) Great, this way we won't invent any version here. You can do it directly in the Github: https://help.github.com/articles/creating-releases/ An example: https://github.com/michael-lazar/rtv/releases Regards. (In reply to Danilo G. Baio from comment #11) Done. I created 3 tags which reflect the release history of the project. The latest release is v1.0.2 which contains changes which I just committed today: The commit message tells all: "FadeMind has removed the windows 10 telemetry hosts in favor of using the respective hosts from WindowsSpyBlocker, so we do this as well." It would be good to submit v1.0.2 to the ports, so users won't be confused with a message from the missing telemetry hosts file from FadeMind. Created attachment 185671 [details]
Patch to update to 1.0.2 (svn diff)
Nice! The attached patch will update the port to v1.0.2.
portlint ok.
poudriere testport ok: 10.3-amd64, 10.3-i386, 11.0-amd64, 11-0-i386, 11.1-amd64, 11.0-arm.
@cyclaero, that's great, thanks. Besides the release version be documented, every time you tag a new version on Github, the maintainer here will be notified about the new version through portscout [1]. @vidar, with attachment #185671 [details] I got this error: > ===> Patching for void-zones-tools-1.0.2 > ===> Applying FreeBSD patches for void-zones-tools-1.0.2 > 1 out of 1 hunks failed--saving rejects to Makefile.rej > => FreeBSD patch patch-Makefile failed to apply cleanly. > *** Error code 1 Removing `files/patch-Makefile` builds fine here: 10.3-RELEASE-p21 [amd64|i386] 11.0-RELEASE-p12 [amd64|i386] 12.0-CURRENT 1200040 [amd64|i386] Could you or @mikael.urankar confirm if it works this way on ARM as well? Thank you. [1] - https://portscout.freebsd.org/ (In reply to Danilo G. Baio from comment #14) Yes, files/patch-Makefile is no longer applicable. I deleted it on my end before I ran the various test builds, but I forgot to mention it here (and the file deletion doesn't show up in the svn diff). A commit references this bug: Author: dbaio Date: Wed Aug 23 07:35:03 UTC 2017 New revision: 448592 URL: https://svnweb.freebsd.org/changeset/ports/448592 Log: dns/void-zones-tools: Update to 1.0.2 - Use Github release - Remove `files/patch-Makefile`, not necessary anymore. - Fixes for building on ARM and i386 architectures - WindowsSpyBlocker is the new upstream for the Windows telemetry hosts files PR: 217054 Submitted by: Vidar Karlsen <vidar@karlsen.tech> (maintainer) Changes: head/dns/void-zones-tools/Makefile head/dns/void-zones-tools/distinfo head/dns/void-zones-tools/files/ A commit references this bug: Author: dbaio Date: Wed Aug 23 07:41:20 UTC 2017 New revision: 50704 URL: https://svnweb.freebsd.org/changeset/doc/50704 Log: Add Vidar Karlsen to contributors Maintainer of: dns/void-zones-tools security/py-onetime sysutils/pcapfix textproc/crunch PR: 217054 Changes: head/en_US.ISO8859-1/articles/contributors/contrib.additional.xml Committed, thank you all. |
Created attachment 179924 [details] svn diff Update to latest upstream commit, which adds some fixes for building on ARM. Tests done: portlint: ok poudriere testport on 10.3-amd64: ok poudriere testport on 11.0-amd64: ok I don't have any ARM hardware so I haven't been able to test that it builds and runs on ARM.