Regex deletes on the keys may be used to delete interconnected groups of
keys, but I think it is not very efficient. Lets assume we have two sets
of data for a DVD: details and actors. The title of a DVD changes, I
want to update all relevant data and not have stale titles on the
plattform. To delete all connections with one DVD I would have to:
- delete the key for details (probably dvd_1234_details) - easy
- delete the keys for all related actors - this is harder. encode all
dvd_ids in the actor keys? Iterate the actors and run a regex for each one?
Point 2 multiplies for each loosely related keys. The direct ones are
easy to hit, the others not so.
Additionally, some people (including us) programmatically generate the
name of the key. We use the service name, request parameters, sort etc.
to build a long string and then md5 that for a key name. We have no way
of knowing what the key names are, other than rebuilding each
combination of parameters. Then we don't need a regex either.
Sorry for my vage explanations, I studied business ;-)
------------------------------
Message: 3
Date: Tue, 17 Jul 2007 20:21:15 +0200
From: "BUSTARRET, Jean-francois" <[EMAIL PROTECTED]>
Subject: RE: Hackathon notes (non-binary protocol thread)
To: <[email protected]>
Message-ID:
<[EMAIL PROTECTED]>
Content-Type: text/plain; charset="iso-8859-1"
I'm not a very experienced C coder, but I'll try to share some ideas... Please
be tolerant
- add "char* tags" to the _stritem struct
- store tags in a comma/whatever-delimited string : ",tag1,tag2,tag3," (space
and comma being reserved), the client being eventually responsible for formatting the tag
string
- modify the protocol to add an optional tag parameter <command name> <key> <flags>
<exptime> <bytes>[ <tags>]\r\n
- add a new deletetag command ?
- delete items using the same rule approach as described by Dustin,
strpos-ing(",tag,") instead of regexp-ing the key
Multi-tag deletes can be split in multiple rules
IMHO, limiting the max number of tags and the max size of a tag would not be a
problem.
JFB
-----Message d'origine-----
De : Paul Querna [mailto:[EMAIL PROTECTED]
Envoyé : mardi 17 juillet 2007 19:01
À : Tobias Lütke
Cc : BUSTARRET, Jean-francois; [email protected]
Objet : Re: Hackathon notes (non-binary protocol thread)
Tobias Lütke wrote:
Agreed,
I think the tagging solves the same problem as the wildcard deletes in
a more elegant way. The concept is so simple that you can explain it
in a single sentence which is a indicator of a good feature.
On 7/16/07, BUSTARRET, Jean-francois <[EMAIL PROTECTED]> wrote:
What about tagging (as discussed last week on the mailing-list) ?
IMHO, it would be more useful than wildcard deletes/namespaces/...
Well, what it came down to is that we had very simple ideas for how wildcard
deletes could be implemented, but no one seemed to have any suggestions how to
best do tagging. Have any ideas for the list?
Thanks,
-Paul
--
Geschäftsführer / Managing Director
Hitflip Media Trading GmbH
Gürzenichstr. 7, 50667 Köln
www.hitflip.de - new: www.hitflip.co.uk
Tel. +49-(0)221-272407-27
Fax. 0221-272407-22 (that's so 1990s)
HRB 59046, Amtsgericht Köln
Geschäftsführer: Andre Alpar, Jan Miczaika, Gerald Schönbucher
