Ian Jackson writes ("Re: Bug#923732: dgit-infrastructure:
--deliberately-include-questionable-history required for every single push"):
> Not really in a position to look at this properly but I think I
> understand now and I think you have indeed identified a bug.
I have spent a while trying to reproduce this. I found that there is
indeed code which is supposed to remove taints when they are
overridden.
It wasn't tested at all. But I wrote a sketch of a test, which I
thought would resemble what happened to this package, and the
mechanism seems to work. See attached.
Can you tell me the exact sequence of events up to the first
unwarranted push rejection, with times where you have them ?
Ian.
#!/bin/bash
set -e
. tests/lib
t-tstunt-parsechangelog
t-debpolicy
t-prep-newpackage example 1.0
cd $p
revision=1
git tag start
t-dgit setup-mergechangelogs
: upload
t-dgit push-source --new
: cut
rm $tmp/incoming/*
t-archive-none example
touch -d 'now -1 day' $tmp/git/example*
t-policy-periodic # xxx maybe want test both with and without this
: edit
t-commit 'edit after cut'
: push, needs --deliberately
t-expect-push-fail \
'all previously pushed versions were found to have been removed from NEW' \
t-dgit push-source --new
t-dgit push-source --new --deliberately-include-questionable-history
t-archive-process-incoming new
: t-policy-periodic # xxx maybe want test both with and without this
: accept
mv -f $tmp/aq/package.{new,sid}.$p
t-aq-archive-updated sid $p
: t-policy-periodic # xxx maybe want test both with and without this
t-git-dir-time-passes
: 3rd push, no deliberately
t-commit 'edit after accept'
t-dgit push-source
t-ok
--
Ian Jackson <ijack...@chiark.greenend.org.uk> These opinions are my own.
If I emailed you from an address @fyvzl.net or @evade.org.uk, that is
a private address which bypasses my fierce spamfilter.
