Invalid tar magic when streaming download

[Jeffrey Fetterman]

Ah. I don't think this mailing list is for me. Thank you for your time.

On Thu, Dec 5, 2019 at 7:31 PM Eli Schwartz <eschwartz at archlinux.org> wrote:
>
> On 12/5/19 8:03 PM, Jeffrey Fetterman wrote:
> > Hey, I'm not the one who decided that both zipballs and tarballs on
> > GitHub should use DEFLATE compression on random files.
> I'm genuinely not sure if you're trolling. The url you used is an HTTP
> 302 redirect to
>
> Location: https://codeload.github.com/mirror/busybox/legacy.tar.gz/master
>
> Which in turn specifies:
>
> Content-Type: application/x-gzip
> Content-Disposition: attachment;
> filename=mirror-busybox-1_31_0-98-g9ec836c.tar.gz
>
> But most people don't bother using the legacy API, and instead use the
> publicly linked urls like
> https://github.com/mirror/busybox/archive/master.tar.gz
>
> Which is a 302 redirect to:
>
> Location: https://codeload.github.com/mirror/busybox/tar.gz/master
>
> Which in turn declares:
>
> Content-Type: application/x-gzip
> Content-Disposition: attachment; filename=busybox-master.tar.gz
>
> So, I'm bewildered why you would be surprised that a compressed tarball
> would be "randomly compressed", and I'm not sure why you are blaming
> github for ruining your experiment or whatever. You want to experiment
> with how tar decompresses files from a pipe? I suggest you use
>
> gzip <infile.tar | busybox tar -xf -
>
> This produce controllable experiments where you know and can account for
> your inputs.
>
> P.S. I'm unaware of anyone hosting software forges, git-based or
> otherwise, who offer non-compressed tarballs for download. It would be
> stupid of them to offer it, because why on earth would you go ahead and
> shoot your own bandwidth in the foot like that?
>
> How did you get into the developer API documentation of github without
> realizing this fact?
>
> --
> Eli Schwartz
> Bug Wrangler and Trusted User
>
> _______________________________________________
> busybox mailing list
> busybox at busybox.net
> http://lists.busybox.net/mailman/listinfo/busybox