Article View: demon.ip.developers
Article #1095Re: Robust design for dosdis expire
From: Chris Croughton
Date: Sat, 27 Apr 1996 00:00
Date: Sat, 27 Apr 1996 00:00
59 lines
2924 bytes
2924 bytes
In article <830215383snz@wash.demon.co.uk>
john@wash.demon.co.uk "John Washington" writes:
>I've just been looking at unbatch.c, having suffered a
>fatal error #12 on Sunday. I didn't fancy using reindex
>and losing the articles-read info. 'active' is updated
>before the new article is appended to the file. I would
>have expected that step to be last, and it would then be
>relatively easy to have a utility that would trim the
>news file and index back to the high article number
>recorded in 'active'.
It exists. LSI (Logical SNEWS Indexer) should be around on the FTP
server. It basically just renumbers the active file as you suggest, and
doesn't spoil the other files. It's not a total cure, but as you say
if it fails you can always do reindex...
>wb> BTW, EXPIRE has an alternative mode of working
>wb> which gives me the screaming heeby-jeebies. It takes a parameter and
>wb> applies that to all groups. e.g. expire 0 will delete all news in all
>wb> groups - even those with -ve expiry times. Opinions on
>wb> changing/removing that feature?
Kill it. It's a left-over from before the expire.dat file was used.
However, something I would find useful would be to have a command-line
option to expire just one newsgroup. Something like
expire demon.mouse.droppings 5 mdrop.bat
(the last parameter being an optional archive file as used in
expire.dat).
>wb> Malcolm has expressed concerns that EXPIRE should not be any slower than
>
>ROTFL.
>
>So use del *.*, it's amazingly efficient.
>
>Users who put speed before accuracy and reliability get what
>they deserve.
I fully agree there. The suggestions for increasing the speed of expire
(detecting when no articles are to be expired and skipping the
newsgroup, for instance) will result in far better time savings and at
no cost in reliability.
>John Washington
Oh, it's you, John. I wondered why I was agreeing so much with the
person I was replying to <g>. I never think to look at the header to
see who wrote the message <g>...
.-------------------------------.-------------------------------------.
| chris@keris.demon.co.uk | FIAWOL (Filking Is A Way Of Life) |
`-------------------------------^-------------------------------------'
This message is void in Germany, China, the United States,
and other places where free speech is prohibited by law.
Technical Details
Message-ID:
<830611160snz@keris.demon.co.uk>Path:
arc1-ca.pugleaf.net!unknown!not-for-mailReferences:
<829700843snz@tsys.demon.co.uk>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: 10bb05,69cce58401bcf88 X-Google-Attributes: gid10bb05,public From: Chris Croughton <chris@keris.demon.co.uk> Subject: Re: Robust design for dosdis expire Date: 1996/04/27 Message-ID: <830611160snz@keris.demon.co.uk> X-Deja-AN: 152118748 x-mail2news-path: keris.demon.co.uk x-nntp-posting-host: keris.demon.co.uk references: <829700843snz@tsys.demon.co.uk> organization: Keristor Systems newsgroups: demon.ip.developers