Agree.
Seems this random access was achieved with a new class TarFile and it won't
have effects on existing apps.
Tars are designed to be a sequence access format so a random access would cost
more time. I'm just curious how much do we want to have a random access here? I
mean, do we always need a random access to a tarball? Is this a reasonable
trade off?
cheers,
Lee
On 11 16 2020, at 8:23 , Gary Gregory <garydgreg...@gmail.com> wrote:
> Reading in everything to allow for random access should be something I have
> to turn on explicitly, using a new parameter/API. IOW, existing apps should
> not have to pay the price for a feature they don't want. Think of an app
> that looks at large tar...
>
> Gary
> On Sun, Nov 15, 2020, 21:51 Peter Lee <peter...@apache.org> wrote:
> > Hi all,
> >
> > Robin has pushed a PR in github in July that implemented the random access
> > of tar :
> > https://github.com/apache/commons-compress/pull/113
> > It's achieved by reading the tar once and have the start position of each
> > tar entry stored. Tar is not designed to be random accessed so this may be
> > a time consuming way.
> > IMO this is not a must feature, but it's a good idea that may be useful
> > when we only want some specific entries in giant tars. This is a big change
> > in tar so I want to hear your voices. WDYT?
> > cheers,
> > Lee
> >
>
