| 1 |
El dom, 11-05-2014 a las 19:46 +0200, Michał Górny escribió: |
| 2 |
> Hello, developers. |
| 3 |
> |
| 4 |
> I'd like to raise the following item for discussion: making .xz |
| 5 |
> the default compressor used by portage for documentation, man pages |
| 6 |
> and info files. That is, the equivalent of: |
| 7 |
> |
| 8 |
> PORTAGE_COMPRESS=xz |
| 9 |
> |
| 10 |
> in make.globals. |
| 11 |
> |
| 12 |
> Rationale: xz-utils is quite widespread nowadays and it is a part |
| 13 |
> of @system set. It can achieve better compression ratio than bzip2, |
| 14 |
> and faster decompression at the same time. |
| 15 |
> |
| 16 |
> I have confirmed that both sys-apps/man and sys-apps/man-db can |
| 17 |
> handle .xz compressed man pages, and sys-apps/texinfo can handle .xz |
| 18 |
> compressed info pages. Major text editors and pagers support .xz |
| 19 |
> alike .bz2 (i.e. usually they support both or neither :)). |
| 20 |
> |
| 21 |
> The additional question is: what preset to use? To help discussing |
| 22 |
> this, I'd like to quote the tables from 'man xz': |
| 23 |
> |
| 24 |
> Preset DictSize CompCPU CompMem DecMem |
| 25 |
> -0 256 KiB 0 3 MiB 1 MiB |
| 26 |
> -1 1 MiB 1 9 MiB 2 MiB |
| 27 |
> -2 2 MiB 2 17 MiB 3 MiB |
| 28 |
> -3 4 MiB 3 32 MiB 5 MiB |
| 29 |
> -4 4 MiB 4 48 MiB 5 MiB |
| 30 |
> -5 8 MiB 5 94 MiB 9 MiB |
| 31 |
> -6 8 MiB 6 94 MiB 9 MiB |
| 32 |
> -7 16 MiB 6 186 MiB 17 MiB |
| 33 |
> -8 32 MiB 6 370 MiB 33 MiB |
| 34 |
> -9 64 MiB 6 674 MiB 65 MiB |
| 35 |
> |
| 36 |
> Preset DictSize CompCPU CompMem DecMem |
| 37 |
> -0e 256 KiB 8 4 MiB 1 MiB |
| 38 |
> -1e 1 MiB 8 13 MiB 2 MiB |
| 39 |
> -2e 2 MiB 8 25 MiB 3 MiB |
| 40 |
> -3e 4 MiB 7 48 MiB 5 MiB |
| 41 |
> -4e 4 MiB 8 48 MiB 5 MiB |
| 42 |
> -5e 8 MiB 7 94 MiB 9 MiB |
| 43 |
> -6e 8 MiB 8 94 MiB 9 MiB |
| 44 |
> -7e 16 MiB 8 186 MiB 17 MiB |
| 45 |
> -8e 32 MiB 8 370 MiB 33 MiB |
| 46 |
> -9e 64 MiB 8 674 MiB 65 MiB |
| 47 |
> |
| 48 |
> I'd like to note here that increasing dictionary size over file size |
| 49 |
> does not improve compression. However, the options involved in CompCPU |
| 50 |
> may. |
| 51 |
> |
| 52 |
> Depending on the expected amount of complexity, I'd either go for: |
| 53 |
> |
| 54 |
> 1) -6e (or -6, the default) -- max CompCPU, reasonable use of memory, |
| 55 |
> and dictionary larger than most (or all?) documents that are going to |
| 56 |
> be compressed, |
| 57 |
> |
| 58 |
> 2) -Ne with minimal 'N' for CompCPU==8 and DictSize > filesize -- still |
| 59 |
> max compression ratio while keeping lowest memory requirements possible. |
| 60 |
> |
| 61 |
> Your thoughts? |
| 62 |
> |
| 63 |
|
| 64 |
Per: |
| 65 |
https://bugs.gentoo.org/show_bug.cgi?id=372653 |
| 66 |
|
| 67 |
Looks like bzip2 was still better for small files :/ |
Archives