Robert Collins wrote:
> On Mon, 2002-12-16 at 14:17, Joe Cooper wrote:
>
>>I thought the null cache dir left out disk access but still kept a
>>memory cache, while no_cache deny all will never cache anything even to
>>cache_mem?
>>
>>Or am I misunderstanding the suggestion or the current behavior?
>
>
> Just me being
> a) brain dead
> b) too brief.
>
> Henrik's later email aligns neatly with my proposal.
>
> The basic thing is, the null cache dir is a workaround for two
> behaviours of squid. Namely squid requires a cache_dir to operate (a),
> and defaults to a ufs cachedir of 100 Mb if none is specified (b).
>
> Addressing (b) is easy, we change the default to be no cache dir.
> (a) will take a careful code review, and thats about all.
>
> Having no null cache dir will allow a somewhat stricter SwapDir class
> definition, and will reduce the number of cache dir code paths to
> maintain to 2 - coss and ufs.
Ah, makes good sense. No arguments from me.
-- Joe Cooper <joe@swelltech.com> Web caching appliances and support. http://www.swelltech.comReceived on Mon Dec 16 2002 - 12:21:31 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:19:01 MST