Current net-analyzer/nfdump-1.6.19: 1) uses incorrect slot version 2) does not allow of NAT events information per upstream 3) unconditionally depends on doxygen (should be conditional) 4) does not actually need the patches 5) does not install generated docs
Pull request with fixes: https://github.com/gentoo/gentoo/pull/14958
(In reply to Vjaceslavs Klimovs from comment #0) > Current net-analyzer/nfdump-1.6.19: > 1) uses incorrect slot version > 2) does not allow of NAT events information per upstream > 3) unconditionally depends on doxygen (should be conditional) > 4) does not actually need the patches > 5) does not install generated docs Please provide actual arguments for all of these.
(In reply to Jeroen Roovers from comment #2) > (In reply to Vjaceslavs Klimovs from comment #0) > > Current net-analyzer/nfdump-1.6.19: > > 1) uses incorrect slot version > > 2) does not allow of NAT events information per upstream [...] > > 4) does not actually need the patches [...] > Please provide actual arguments for all of these. Actually just those. The documentation fixes are easy.
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=9ba2635768ec1b941ece310c1496c07cc5792843 commit 9ba2635768ec1b941ece310c1496c07cc5792843 Author: Jeroen Roovers <jer@gentoo.org> AuthorDate: 2020-08-02 10:19:02 +0000 Commit: Jeroen Roovers <jer@gentoo.org> CommitDate: 2020-08-02 10:43:42 +0000 net-analyzer/nfdump: Version 1.6.21 Package-Manager: Portage-3.0.1, Repoman-2.3.23 Bug: https://bugs.gentoo.org/712590 Signed-off-by: Jeroen Roovers <jer@gentoo.org> net-analyzer/nfdump/Manifest | 1 + net-analyzer/nfdump/nfdump-1.6.21.ebuild | 71 ++++++++++++++++++++++++++++++++ 2 files changed, 72 insertions(+)
Thanks for fixing some of the things! I've synced with your updated ebuild uploaded my version of the ebuild at https://github.com/vklimovs/portage-overlay/blob/master/net-analyzer/nfdump/nfdump-1.6.21-r1.ebuild. Remaining differences, that I believe are for the better are: 1) SLOT version? I could be wrong here, but it appears to be mirroring version, so should be 1.6.21. 2) No support for jnat and nsel functionality in Gentoo's ebuild, this has no dependencies and is surely useful to many folks 3) ftconv does not actually depend on zlib 4) nfpcapd use flag builds a useful daemon and has as simple dependency 5) to actually build docs, doxygen needs to be built with dot (I tried) 6) Everything builds and works as expected without the patches, except for one line which i replaced with sed 7) I have added init scripts These are small differences, but they do make things better and certainly not worth without adding any maintenance burden. I am very happy to create a pull request, please let me know.
(In reply to Vjaceslavs Klimovs from comment #5) > Remaining differences, that I believe are for the better are: > 1) SLOT version? I could be wrong here, but it appears to be mirroring > version, so should be 1.6.21. No, that's absolutely wrong. The sub-SLOT mirrors the last SONAME change, not anything arbitrary like a package's version. > 5) to actually build docs, doxygen needs to be built with dot (I tried) Bug #430914. If nfdump requires media-gfx/graphviz to build documentation it should state so and not depend on a USE flag on doxygen that merely pulls in media-gfx/graphviz as dependency. > 6) Everything builds and works as expected without the patches, except for > one line which i replaced with sed That's not an "issue"? That things just "happen" to work for you does not mean the patches are useless. Why would you replace a patch with a sed script? The -compiler patch cleans up CFLAGS injections by the build system. This is a thing that we do at Gentoo Linux. The -libft patch replaces some checks on some paths to things that might be libraries to the autoconf recommended way of checking whether a library resolves a specific symbol. Again, not using that patch might happen to work for you, but that does not argue against its use. > These are small differences, but they do make things better and certainly > not worth without adding any maintenance burden. All items I did not respond to should be good to propose changes for. Perhaps you could put precisely those changes in an ebuild? > I am very happy to create a pull request, please let me know. You did that already, I think?
Thanks for your explanations. Here is a pull request with small remaining improvements: https://github.com/gentoo/gentoo/pull/17119
Any news? nsel and jnat use flags are will be very useful
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=6dace4048ed813d3811cdb8d325fd4519c117753 commit 6dace4048ed813d3811cdb8d325fd4519c117753 Author: Vjaceslavs Klimovs <vklimovs@gmail.com> AuthorDate: 2020-08-14 04:32:18 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2021-03-15 21:03:37 +0000 net-analyzer/nfdump: add use flags and init scripts for nfcapd Closes: https://bugs.gentoo.org/712590 Signed-off-by: Vjaceslavs Klimovs <vklimovs@gmail.com> Signed-off-by: Sam James <sam@gentoo.org> net-analyzer/nfdump/files/nfcapd.confd | 11 ++++ net-analyzer/nfdump/files/nfcapd.initd | 60 ++++++++++++++++++++++ net-analyzer/nfdump/metadata.xml | 8 ++- net-analyzer/nfdump/nfdump-1.6.21-r1.ebuild | 78 +++++++++++++++++++++++++++++ 4 files changed, 156 insertions(+), 1 deletion(-)
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=7fad0713ae0b090bea16187f8b4d546dd1106b1f commit 7fad0713ae0b090bea16187f8b4d546dd1106b1f Author: Sam James <sam@gentoo.org> AuthorDate: 2021-03-15 21:10:20 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2021-03-15 21:10:20 +0000 net-analyzer/nfdump: apply patch to 1.6.21 Bug: https://bugs.gentoo.org/712590 Signed-off-by: Sam James <sam@gentoo.org> net-analyzer/nfdump/{nfdump-1.6.21.ebuild => nfdump-1.6.21-r1.ebuild} | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
