|WikiProject Computing||(Rated Start-class, Low-importance)|
|WikiProject Linux||(Rated Start-class)|
The disadvantages section
I removed this, because it is demonstrably false:
- The cloop kernel module only allows one compressed image at a time to be active.
Although you are only allowed to specify one compressed image on the insmod cloop.o command line, you can use the losetup utility to activate multiple compressed images.
I'm not sure how true the remaining two entries are.
Also, is there any reason to have a "disadvantages" section but no "advantages" section? It seems kind of out-of-place. 126.96.36.199 17:17, 1 October 2005 (UTC)
New version of software and new file format?
The description on this page, information on the linked pages, and information that can be found by search the web seems to describe an old version of cloop. For example, cloop 2.04-118 that comes with OpenSuSE 10.3 behaves very different:
- multiple images can be allowed by creating additional device nodes (mknod /dev/cloop$I b 240 $I, I=0,...,7, and losetup, see also discussion section above)
- create_compressed_fs does not hold the compressed data in (virtual) memory if the image is a regular file. Instead, it starts outputting data immediately (tested with a 55 GB image file). This also means that the file format must have changed since the index cannot be written at the start of the output anymore (stdout is not seekable). Jowagner (talk • contribs) 17 January 2008 (UTC)
Claimed slowness as disadvantage
The article says "The design of the cloop driver requires that compressed blocks be read whole from disk. This makes cloop access very slow when there are many small scattered reads"
In my own testing, cloop has consistently *sped up* operations on small reads. I can't reproduce the above claim at all. See my detailed results on: http://www-etud.iro.umontreal.ca/~jaegerch/div/cloop_performance_tests —Preceding unsigned comment added by Pflanze2 (talk • contribs) 18:59, 14 February 2010 (UTC)
I've added more testing, with a blocksize of 256KB (4 times the default), and can see some slowdown in one case when the CPU is artificially slowed down. I think you still need a fast disk to see the slowdown at all. Anyway the section still needs some rewording and a citation. —Preceding unsigned comment added by Pflanze2 (talk • contribs) 19:44, 14 February 2010 (UTC)