Bug#918438: orig tarball components with uppercase letters
Andrey Rahmatullin writes ("Re: Bug#918438: orig tarball components with uppercase letters"):
> From these two at least VFAT, while case-insensitive, is case-preserving.
> ISO9660 without both Rock Ridge and Joliet is, I suspect, already
Err, I think you're right about ISO9660 without either Rock Ridge or
Joliet. But ISO9660 with Joliet but not RR is case-insensitive and
The main problem is that case-insensitive but case-preserving is not
sufficient, since the current spec makes it possible for two different
files to have names which are the same when case-smashed.
So what I think is needed is to come up with some rule which
(i) prevents two different files with case-insensitively-identical
names (ii) can actually somehow be assured to be true.
The simplest such rule is to forbid uppercase letters completely. It
is not necessarily the most convenient now because that involves a
change, but it is a small change which should be easy to incorprate
into everyone's workflow eventually, and I think we aren't in a hurry
because this is right now a theoretical problem.
That is the only rule violations of which can be spotted locally,
without some kind of context or database or something.
More complex rules are possible. For example, (i) no single package
may contain multiple case-insensitively-identical orig components plus
(ii) a similar condition on different uploads of the same (package,
orig-version). The condition (ii) would have to be clearly stated and
then implemented in dak and reprepro and launchpad (at least).
Does that make sense ?
Ian Jackson <email@example.com> 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.