Mailing-List: contact dev-help@apr.apache.org; run by ezmlm
Precedence: bulk
Date: Sat, 08 Sep 2001 13:51:06 -0700
From: Brian Pane <bpane@pacbell.net>
Subject: Re: [PATCH] Turn apr_table_t into a hash table
To: rbb@covalent.net
Cc: dev@httpd.apache.org, dev@apr.apache.org
Message-id: <3B9A84BA.7070607@pacbell.net>
MIME-version: 1.0
Content-type: text/plain; charset=us-ascii; format=flowed
Content-transfer-encoding: 7BIT
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.3) Gecko/20010801
References: <3B993AEB.8040602@pacbell.net>
 <20010908153211.A49E046DF4@koj.rkbloom.net> <3B9A6371.1000204@pacbell.net>
 <20010908202001.2EE4646DF4@koj.rkbloom.net>

Ryan Bloom wrote:

>On Saturday 08 September 2001 11:29, Brian Pane wrote:
>
>>Ryan Bloom wrote:
>>
>>>On Friday 07 September 2001 14:23, Brian Pane wrote:
>>>
>>>>The attached patches change the apr_table_t implementation from
>>>>a linear list to a hash table (not an apr_hash_t, though!).  With
>>>>this change, I'm seeing a ~3% improvement in throughput when
>>>>delivering a 0-byte file over the loopback on Linux.  (I used this
>>>>0-byte test case to measure the inherent overhead in the httpd, without
>>>>transmission time clouding the results.)
>>>>
>>>I dislike this.  Why are we putting a second hash table into APR?  If we
>>>want to use a hash, then ues an apr_hash_t.  If apr_hash_t doesn't
>>>support something that we MUST have to do this, then fix apr_hash_t. 
>>>Having two different hash alorithms in APR, one of them hidden under a
>>>tables API, seems kind of hackish to me.
>>>
>>Are you arguing in favor of using apr_hash_t in the implementation of
>>apr_table_t,
>>or using apr_hash_t in place of apr_table_t in the request_rec?  I'm
>>comfortable
>>with the former, but not the latter.
>>
>
>The latter.  Having two API's to the same functions should only be done
>as a stop-gap.
>
I disagree.  It's inevitable to have two APIs, as long as we have two
'table' types with very different semantics.

apr_table_t is statically typed (uses char*), and apr_hash_t isn't (uses 
void*).
If we did a direct replacement of apr_table_t with apr_hash_t in the httpd,
we'd be sacrificing static type-safety for a lot of code.

>  Also, a LOT of modules using the apr_table_elem macro
>to get the elements from a table.  Turning that into an API would break a
>lot of people.
>
My patch covers all the changes needed for the core modules.  Do you have
any info on the number of 3rd-party modules for 2.0 that use apr_table_elem?

--Brian