| 1 |
Source: libcache-fastmmap-perl |
| 2 |
Standards-Version: 4.7.4 |
| 3 |
Maintainer: Debian Perl Group <[email protected]> |
| 4 |
Uploaders: |
| 5 |
gregor herrmann <[email protected]>, |
| 6 |
Ansgar Burchardt <[email protected]>, |
| 7 |
Harlan Lieberman-Berg <[email protected]>, |
| 8 |
Section: perl |
| 9 |
Testsuite: autopkgtest-pkg-perl |
| 10 |
Build-Depends: |
| 11 |
debhelper-compat (= 13), |
| 12 |
libcompress-lz4-perl <!nocheck>, |
| 13 |
libcompress-snappy-perl <!nocheck>, |
| 14 |
libtest-deep-perl <!nocheck>, |
| 15 |
perl-xs-dev, |
| 16 |
perl:native, |
| 17 |
Vcs-Browser: https://salsa.debian.org/perl-team/modules/packages/libcache-fastmmap-perl |
| 18 |
Vcs-Git: https://salsa.debian.org/perl-team/modules/packages/libcache-fastmmap-perl.git |
| 19 |
Homepage: https://metacpan.org/release/Cache-FastMmap |
| 20 |
|
| 21 |
Package: libcache-fastmmap-perl |
| 22 |
Architecture: any |
| 23 |
Depends: |
| 24 |
${misc:Depends}, |
| 25 |
${perl:Depends}, |
| 26 |
${shlibs:Depends}, |
| 27 |
libcompress-lz4-perl, |
| 28 |
libcompress-snappy-perl, |
| 29 |
Description: Perl module providing a mmap'ed cache |
| 30 |
Cache::FastMmap uses the mmap system call to establish an interprocess shared |
| 31 |
memory cache. Its core code is written in C, which can provide significant |
| 32 |
performance compared to a Pure Perl implementation such as Cache::Mmap. It can |
| 33 |
handle rather large pages without the socket connection and latency of using |
| 34 |
full-fledged databases where long-term persistence is unnecessary. |
| 35 |
. |
| 36 |
Since the algorithm uses a dual-level hashing system (a hash is used to find |
| 37 |
a page, then another hash within each page to find a given slot), most get |
| 38 |
calls can execute in constant O(1) time. The system uses fcntl to handle |
| 39 |
concurrent access, but only locks individual pages to reduce contention. The |
| 40 |
oldest (least recently used) data is evicted from the cache first, making |
| 41 |
this cache implementation most suitable for cases when old data is unlikely |
| 42 |
to be searched. |