Henning,
The memcached module is very useful. But I think it would be about 10x more useful if expiration values could be controlled on a per-key basis rather than just globally as a modparam, because different kinds of data needs to be aged out at different intervals.
What do you think?
-- Alex
On Freitag, 2. Oktober 2009, Alex Balashov wrote:
The memcached module is very useful. But I think it would be about 10x more useful if expiration values could be controlled on a per-key basis rather than just globally as a modparam, because different kinds of data needs to be aged out at different intervals.
Hello Alex,
this is indeed a good suggestion. I've implemented some functionality that implements this like the 'htable' module, e.g. added a new PV "$mctex(key) = expire time" to change the expiration time for an existing value. Does this matches your idea?
I want to tests this a bit before committing, hope to finish this today or tomorrow.
Regards,
Henning
Henning Westerholt wrote:
this is indeed a good suggestion. I've implemented some functionality that implements this like the 'htable' module, e.g. added a new PV "$mctex(key) = expire time" to change the expiration time for an existing value. Does this matches your idea?
Yes.
Wow, thank you Henning! I did not expect you to actually implement it; I was just throwing it out as a suggestion. But this is wonderful and will certainly transform the situation. I really appreciate it.
On Montag, 5. Oktober 2009, Alex Balashov wrote:
this is indeed a good suggestion. I've implemented some functionality that implements this like the 'htable' module, e.g. added a new PV "$mctex(key) = expire time" to change the expiration time for an existing value. Does this matches your idea?
Yes.
Wow, thank you Henning! I did not expect you to actually implement it; I was just throwing it out as a suggestion. But this is wonderful and will certainly transform the situation. I really appreciate it.
Hey Aley,
ok, then i see that i get the code fast out that you can test it. ;) It was actually not that much work, i was able to combine it with some nice refactorings, and also some mem leak fixes in error conditions.
Regards,
Henning
On Dienstag, 6. Oktober 2009, Henning Westerholt wrote:
Wow, thank you Henning! I did not expect you to actually implement it; I was just throwing it out as a suggestion. But this is wonderful and will certainly transform the situation. I really appreciate it.
ok, then i see that i get the code fast out that you can test it. ;) It was actually not that much work, i was able to combine it with some nice refactorings, and also some mem leak fixes in error conditions.
Hi Alex,
i've commited a initial implementation. I've done some first tests, so far the server seems to get the correct setting (according its debug log). But it seems that with my (minimal) load the server will not expire old entries, so they can be retrieved nevertheless. Perhaps you can give it a try as well.
Cheers,
Henning
-
Alex Balashov -
Henning Westerholt