On Thu, Apr 29, 2021 at 1:36 AM Luke Diamand <luke@xxxxxxxxxxx> wrote: > > I think Andrew Oakley pointed this out earlier - but in the days of > Python2 this was (I think) never a problem. Python2 just took in the > binary data, in whatever encoding, and passed it untouched on to git, > which in turn just stored it. > Unfortunately, I just became aware yesterday that Andrew Oakley was also working on this issue (his CC to me 2 weeks ago somehow ended up in my Spam folder, and I only dug it out of there after finding out that we both created a test with the same number). When I first became aware of Andrew's work (yesterday), I thought it would make mine unnecessary, but upon further investigation, I don't think Andrew's work will solve this problem. Please see my reply yesterday to Andew's thread: https://lore.kernel.org/git/CAKu1iLXRrsB4mRsDfhBH5aahWzDjpfqLuWP9t47RMB=RdpL1iA@xxxxxxxxxxxxxx
