Re: compressing articles not working correctly

[lee <lee@yun.yagibdah.de>]

asjo@koldfront.dk (Adam Sjøgren) writes:

> On Thu, 14 Jul 2011 17:46:37 +0200, lee wrote:
>
>> ,---- [ lee@yun:~/Mail/mail/lists/users-global-libreoffice$ ls -lat ]
>> | -rw-------  1 lee lee   5012 14. Jul 17:26 1161
>> | -rw-------  1 lee lee   4896 14. Jul 17:26 1160
>> | -rw-------  1 lee lee   4909 14. Jul 17:26 1159.bz2
>> `----
>
> ls -l doesn't report the size of the body of the email. The
> documentation of nnml-compressed-files-size-threshold indicates that the
> headers are not included when applying the threshold:
>
> ,----[ C-h v nnml-compressed-files-size-threshold RET ]
> | Default size threshold for compressed message files.
> | Message files with bodies larger than that many characters will
>                      ^^^^^^
> | be automatically compressed if `nnml-use-compressed-files' is
> | non-nil.
> `----
>
> Could that be it?

Hm, looks like it.  I checked above articles and found that the body is
only about 1kB.

The idea is to save disk space without losing more speed than necessary
due to compression.  Each article will occupy at least 4kB because
that's the sector size.  I could probably set the threshold to -1 and
have all articles compressed, but for those articles that are smaller
than 4kB, including headers, compression isn't needed.  The same goes
for articles >4kB<8kB uncompressed and still >4kB<8kB when
compressed. Looks like I should write some program that scans the
article storage and compresses all uncompressed articles depending on
whether the compression does save disk space or not ... The speed gain
might be not worth it because it doesn't take awfully long to
(un-)compress small articles.  Hmm.