Re: [PATCH reproducible-notes] Remove non-sense wireguard note
Hi Ted, On Sun, Jan 26, 2020 at 4:54 AM Theodore Y. Ts'o wrote: > So it's stupid stuff like the choice of compilers and CFLAGS At this point, wireguard-tools package is reproducible actually. At some point it wasn't, due to some older versions (but not all older versions!) of make(1) passing GLOB_NOSORT to glob(3). The most recent version of make fixes this error, and the wireguard package itself just passes some variable through the internal make $(sort) function to address older make versions. Jason
Re: [PATCH reproducible-notes] Remove non-sense wireguard note
On Tue, Jan 21, 2020 at 05:15:16PM +0100, Jason A. Donenfeld wrote: > The comment itself doesn't indicate to me (upstream) much at all, and > a pretty ordinary attempt to figure out what it means didn't yield > much Hi Jason, At least in my experience, most of the time when there are reproducible build problem, it's much more likely to be something for the Debian maintainer to fix. For example, LTO builds[1] do not mix as far as reproducible builds are concerned. And given some potential code generation bugs with LTO which apparently the GCC maintainers weren't interested in addressing[2], I ultimately decided to kill the use of LTO when building e2fsprogs, since it's generally not CPU bound. Improving e2fsck times by a second or two wasn't worth dealing with user bug reports caused by compiler bugs/mischief. [1] https://lists.debian.org/debian-devel/2019/07/msg00606.html [2] https://lists.debian.org/debian-devel/2019/07/msg00610.html So it's stupid stuff like the choice of compilers and CFLAGS --- and that's much more of a packaging issue than an upstream issue. It *is* possible if you're doing something baroque with embedding timestamps in generated files, but in general I suspect it's better to let the Debian package maintainer figure out any issues, and let them send patches back to you as the upstream maintainer if necessary. Cheers, - Ted
Re: Re: [PATCH reproducible-notes] Remove non-sense wireguard note
On Wed, Jan 22, 2020 at 10:01:35AM +0500, Alexander E. Patrakov wrote: > Unfortunately, this message is still non-ideal, because it contains a dead > link. I left the dead link as as such it still contained useful information, while removing the link would have removed that info. (And now the issue has been resolved upstream anyway, which https://tests.reproducible-builds.org/debian/wireguard shows.) -- cheers, Holger --- holger@(debian|reproducible-builds|layer-acht).org PGP fingerprint: B8BF 5413 7B09 D35C F026 FE9D 091A B856 069A AA1C signature.asc Description: PGP signature
Re: Re: [PATCH reproducible-notes] Remove non-sense wireguard note
(got a "550 5.6.0 improper use of 8-bit data in message header", resending without S-MIME signature, sorry for the duplicate) Holger Levsen wrote: I've improve it like this now: $ git log -p -1 commit 172f203eab628bd5df0106b33153dc428d12dd5c Author: Holger Levsen Date: Tue Jan 21 18:07:14 2020 +0100 improve note on wireguard Signed-off-by: Holger Levsen diff --git a/packages.yml b/packages.yml index 748c9dda..39266770 100644 --- a/packages.yml +++ b/packages.yml @@ -27904,7 +27904,8 @@ wipe: wireguard: version: 0.0.20171221-2 comments: | -Curious ordering of every - see https://sources.debian.org/src/wireguard/0.0.20171221-2/src/tools/show.c/#L173-L179 +Unreproducible on 32bit archs only +might be related to https://sources.debian.org/src/wireguard/0.0.20171221-2/src/tools/show.c/#L173-L179 Unfortunately, this message is still non-ideal, because it contains a dead link. According to the git history at https://git.zx2c4.com/wireguard-tools/tree/src/show.c?id=9207dec08fdf6d1ea1f740bd98cf122603e919bd (i.e. at approximately the same date), the highlighted region matches the whole body of the "every()" function. Wouldn't it make more sense to refer to it instead of giving a dead link? P.S. for me, it is still not obvious how this function, or whatever gets inlined into it, would lead to a non-reproducible build. -- Alexander E. Patrakov
Re: [PATCH reproducible-notes] Remove non-sense wireguard note
Hi Jason, thanks for reaching out to us! On Tue, Jan 21, 2020 at 05:15:16PM +0100, Jason A. Donenfeld wrote: > I received a reply about not providing "private support" I believe this is some unfortunate wording from someone to busy. I believe it was meant to say "please send this request to a mailinglist so others can chime in". the mailing list in question would have been reproducible-bui...@lists.alioth.debian.org which is slightly better than debian-devel@l.d.o. But anyway... > and given a > place to comment online. When I clicked that, I was then required to > make a "-guest" account using my email address on a website. Seemed a > bit high overhead. we are also happy to take email patches or just comments... > The comment itself doesn't indicate to me (upstream) much at all, and > a pretty ordinary attempt to figure out what it means didn't yield > much. Seems like removal is in good order. I've improve it like this now: $ git log -p -1 commit 172f203eab628bd5df0106b33153dc428d12dd5c Author: Holger Levsen Date: Tue Jan 21 18:07:14 2020 +0100 improve note on wireguard Signed-off-by: Holger Levsen diff --git a/packages.yml b/packages.yml index 748c9dda..39266770 100644 --- a/packages.yml +++ b/packages.yml @@ -27904,7 +27904,8 @@ wipe: wireguard: version: 0.0.20171221-2 comments: | -Curious ordering of every - see https://sources.debian.org/src/wireguard/0.0.20171221-2/src/tools/show.c/#L173-L179 +Unreproducible on 32bit archs only +might be related to https://sources.debian.org/src/wireguard/0.0.20171221-2/src/tools/show.c/#L173-L179 -- cheers, Holger --- holger@(debian|reproducible-builds|layer-acht).org PGP fingerprint: B8BF 5413 7B09 D35C F026 FE9D 091A B856 069A AA1C signature.asc Description: PGP signature
Re: [PATCH reproducible-notes] Remove non-sense wireguard note
On Tue, Jan 21, 2020 at 5:25 PM Sam Hartman wrote: > > > "Jonathan" == Jonathan Carter writes: > > Jonathan> On 2020/01/21 16:43, Jason A. Donenfeld wrote: > >> This note doesn't make sense. It's either entirely invalid or so poorly > >> written that it's useless. As the author of the code in question, I've > >> been unable to ascertain what the note is about, and an email to the > note > >> author hasn't yielded any understanding. > > Jonathan> Well, if that's the way you approach them then you shouldn't at > all be > Jonathan> surprised if you don't get a response. > > It might promote a better community to provide constructive advice > rather than just telling people they are doing a bad job. > > To that end, the original commit message might be improved by saying > something like "I could not understand the note and didn't get > clarification when I reached out to the author." rather than including > worgs like "completely wrong" or "useless". Okay. Anyway, here's a patch. I suppose you can now choose to commit this to the repo, justify the existence of the lines it removes, or simply ignore this email thread entirely.
Re: [PATCH reproducible-notes] Remove non-sense wireguard note
On Tue, Jan 21, 2020 at 5:10 PM Jonathan Carter wrote: > > On 2020/01/21 16:43, Jason A. Donenfeld wrote: > > This note doesn't make sense. It's either entirely invalid or so poorly > > written that it's useless. As the author of the code in question, I've > > been unable to ascertain what the note is about, and an email to the note > > author hasn't yielded any understanding. > > Well, if that's the way you approach them then you shouldn't at all be > surprised if you don't get a response. My approach to it was actually: """ Hi Chris, I noticed you added the following comment to the reproducible-notes repo. Why? What is the significance of this? How does it relate to reproducible builds? What is your observation about this code exactly? Thanks, Jason """ I received a reply about not providing "private support" and given a place to comment online. When I clicked that, I was then required to make a "-guest" account using my email address on a website. Seemed a bit high overhead. The comment itself doesn't indicate to me (upstream) much at all, and a pretty ordinary attempt to figure out what it means didn't yield much. Seems like removal is in good order. Jason
Re: [PATCH reproducible-notes] Remove non-sense wireguard note
> "Jonathan" == Jonathan Carter writes: Jonathan> On 2020/01/21 16:43, Jason A. Donenfeld wrote: >> This note doesn't make sense. It's either entirely invalid or so poorly >> written that it's useless. As the author of the code in question, I've >> been unable to ascertain what the note is about, and an email to the note >> author hasn't yielded any understanding. Jonathan> Well, if that's the way you approach them then you shouldn't at all be Jonathan> surprised if you don't get a response. It might promote a better community to provide constructive advice rather than just telling people they are doing a bad job. To that end, the original commit message might be improved by saying something like "I could not understand the note and didn't get clarification when I reached out to the author." rather than including worgs like "completely wrong" or "useless". --Sam
Re: [PATCH reproducible-notes] Remove non-sense wireguard note
On 2020/01/21 16:43, Jason A. Donenfeld wrote: > This note doesn't make sense. It's either entirely invalid or so poorly > written that it's useless. As the author of the code in question, I've > been unable to ascertain what the note is about, and an email to the note > author hasn't yielded any understanding. Well, if that's the way you approach them then you shouldn't at all be surprised if you don't get a response. -Jonathan -- ⢀⣴⠾⠻⢶⣦⠀ Jonathan Carter (highvoltage) ⣾⠁⢠⠒⠀⣿⡁ https://wiki.debian.org/highvoltage ⢿⡄⠘⠷⠚⠋ https://debian.org | https://jonathancarter.org ⠈⠳⣄ Be Bold. Be brave. Debian has got your back.
[PATCH reproducible-notes] Remove non-sense wireguard note
This note doesn't make sense. It's either entirely invalid or so poorly written that it's useless. As the author of the code in question, I've been unable to ascertain what the note is about, and an email to the note author hasn't yielded any understanding. Signed-off-by: Jason A. Donenfeld --- packages.yml | 4 1 file changed, 4 deletions(-) diff --git a/packages.yml b/packages.yml index 748c9dda..1b93a437 100644 --- a/packages.yml +++ b/packages.yml @@ -27901,10 +27901,6 @@ wipe: version: 0.24-2 issues: - gcc_captures_build_path -wireguard: - version: 0.0.20171221-2 - comments: | -Curious ordering of every - see https://sources.debian.org/src/wireguard/0.0.20171221-2/src/tools/show.c/#L173-L179 wireshark: version: 1.12.7+g7fc8978-1 comments: | -- 2.24.1