Ah, I understand the status of this issue (that will do for me), and I agree w/
every point you mentioned.
Wez Furlong <Pine.LNX.4.53.0305011157120.7207@zaneeb.brainnet.i>:
>I am aware of this issue, however, reverting leaves us at the mercy of
>problems on solaris.
>
>I do plan to implement a write buffer, however, introducing that into
>the 4.3.x branch is probably a bad idea at this late stage.
>
>While I don't think that stream_set_write_buffer()/set_file_buffer()
>having no effect is such a bad thing (most php scripts don't have any
>need to call this function), I am concerned about the overall
>performance of writes in the absence of a user-space write buffer.
>
>So I'm looking into a solution at the moment; there is no need to update
>the documentation at this point.
>
>--Wez.
>
>
>On Thu, 1 May 2003, Masaki Fujimoto wrote:
>
>> Hello,
>>
>> I noticed that stream_set_write_buffer() is no more available on local files
>> because of the change made on 2003/03/18 to main/streams.c (1.125.2.41 ->
>> 1.125.2.42: use a file descriptor instead of a file pointer).
>>
>> As a result of this change, of course stream_set_write_buffer() and
>> set_file_buffer() does not work anymore in most of the cases because user-level
>> I/O buffer functions like setvbuf() does not work on file descriptors, though
>> I'm not sure if this is the "probelm" or not.
>>
>> If this is an intentional result, fixing documents (== add note like "stream_set
>> _write_buffer does not work on local files, bla bla bla" and fix a sample code)
>> will do, but if not, maybe we should implement php-level I/O buffer or revert
>> this change (I'm not sure reverting this change makes another problem,
>> though:)
>>
>> Regards,
>>
>> Masaki Fujimoto
>> fujimoto@tunebiz.net
>> fujimoto@php.net
>>
>>
>>
--
Masaki Fujimoto
fujimoto@techstyle.jp / fujimoto@php.net
http://techstyle.jp/