[Gnu-arch-users] Re: sha1 support

[Colin Walters]

On Sun, 2004-01-18 at 19:01, Tom Lord wrote:

> That's true but it will only reduce, (probably) not eliminate those
> effected by the discontinuity.

True.  But given the community's awareness, I think it would come quite
close to eliminating it.

> Is there a rush for sha1 stuff?   Is there some reason why buffering
> the checksums before writing the file is a bad idea?

It makes the code uglier, and moreover it's code that we'll be stuck
with forever.

> Very often, on #arch at least, people are told to use head or the
> latest preX release because, in spite of the names, they have proven
> to be stable and reliable over a long time period.  Why undermine
> that?

Because it creates an unreasonable expectation that things will never
change incompatibly in a prerelease.  So far that's been true - but the
previous checksum parser, had it been even slightly different, could
very easily have made adding sha1 impossible.

Anyways, I've made the change you requested.  I still don't like it
though.

address@hidden
  tla
    tla--mainline
      tla--mainline--1.2
[...]
        patch-4
          add sha1 lines at end of checksum file

signature.asc
Description: This is a digitally signed message part