Any testers for proposed opportunistic filesystem compression?

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Any testers for proposed opportunistic filesystem compression?

Eric A. Borisch-2
https://github.com/macports/macports-base/pull/76

During the extract-from-package step, if bsdtar exists **and** supports --hfsCompression, use the flag during the extraction.

To test:
 * make sure libarchive is installed
 * check disk usage
 * deactivate & re-activate a large port (llvm / gcc / clang)
 * check disk usage again. If everything is working, you should have more space available.

I've been using it for months and have been very happy with the space savings (especially on packages with lots of text files. *cough* compilers *cough*.)

It is slower to extract, but I don't notice any slowdown after the that point. Most Apple-installed tools are compressed, from the spot-checking I’ve done.

Downsides:
 * Doesn't detect if the compression will be lost (and therefore pointless work/time) when moving for activation. This is the case if different mountpoints are involved for the extraction directory and the final activation locations. Scrapped detection of this as corner case with significantly increased complexity. (And it’s not broken in the corner case, just burns some extra cycles during extraction.)
 * Slower to extract than when not in use -- shouldn't be a surprise.

Thanks,
  - Eric
Reply | Threaded
Open this post in threaded view
|

Re: Any testers for proposed opportunistic filesystem compression?

Jan Stary
On Apr 12 22:58:54, [hidden email] wrote:
> https://github.com/macports/macports-base/pull/76
>  * Slower to extract than when not in use -- shouldn't be a surprise.

Yeah, about six times slower. Please don't.
https://github.com/macports/macports-base/pull/76#issuecomment-382318841

        Jan

Reply | Threaded
Open this post in threaded view
|

Re: Any testers for proposed opportunistic filesystem compression?

Eric A. Borisch-2
See my comments on the thread. Yes, it is slower, but for users with fixed space, the time during activation isn't the biggest concern.

On Wed, Apr 18, 2018 at 4:09 AM, Jan Stary <[hidden email]> wrote:
On Apr 12 22:58:54, [hidden email] wrote:
> https://github.com/macports/macports-base/pull/76
>  * Slower to extract than when not in use -- shouldn't be a surprise.

Yeah, about six times slower. Please don't.
https://github.com/macports/macports-base/pull/76#issuecomment-382318841

        Jan