Hey Lutz (et al),

On Tuesday 22 Oct 2002 3:11 am, Lutz Jaenicke wrote:
> The current behaviour is, that NO_INTERNAL_LOOKUP does not longer cache
> sessions internally since 0.9.6d. Nobody ever complained about this
> behavioural change. (Well it seems that the change I made at that time
> was incomplete and is to be fixed...)

Yeah - I'm not surprised, in actuality I suspect that everyone using
NO_INTERNAL_LOOKUP is using it because they have their own cache
implementation (and callbacks) - ie. they don't notice or care whether
builtin caching was being used as well, they just care that lookups
aren't made in the builtin cache. Well, that is, nobody noticed or cared
until Nadav started looking into it :-)

> With respect to your discussion: if the extension to the API you
> proposed here is required/useful for at least one major application
> (Apache2/mod_ssl) I have no problems to:
> * revert the change in 0.9.6, such that NO_INTERNAL_LOOKUP does again
>   store sessions into the cache,
> * fix the reported bug for the 0.9.7 (and later) tree,
> * add the proposed API extension.
> I would propose to make the API change already to the 0.9.7 tree, as I
> think that the change is small enough and making the behaviour
> consistent seems like a bugfix to me.

<aol>me too</aol>

I think there are a couple of facts arguing for this;
 - nobody currently notices (it seems) that NO_INTERNAL_LOOKUP used to
   store sessions, and now it *doesn't* store sessions except when it
   does (ie. it's gone from contradicting the man page to inconsistently
   contradicting itself).
 - NO_INTERNAL_LOOKUP and NO_INTERNAL_STORE would be logical and
   complimentary flags to implement, and would AFAICS provide
   applications with all the logically useful possibilities of builtin
   caching.
 - Apache2/mod_ssl requires the bitwise OR of NO_INTERNAL_LOOKUP and
   NO_INTERNAL_STORE (I propose "NO_INTERNAL") and currently the
   appropriate flag has never existed (NO_INTERNAL_LOOKUP used to
   implement what I suggest it should, and now implements a buggy mix of
   what it used to implement and what Apache2/mod_ssl needs).

I'll try and put a patch together later for review (and perhaps Nadav
you would be able to grind it through your existing tests?).

Cheers,
Geoff

-- 
Geoff Thorpe
[EMAIL PROTECTED]
http://www.geoffthorpe.net/


______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       [EMAIL PROTECTED]
Automated List Manager                           [EMAIL PROTECTED]

Reply via email to