Justin Erenkrantz wrote:
> On 2/11/06, Peter N. Lundblad <peter@famlundblad.se> wrote:
>
>> I just checked out a copy of trunk from the svn repository. That produced
>> 1367 lines of output and strace reports 123426 syscalls (that's the number
>> of lines output by strace). So, in the case where fflush *is* an extra
>> syscall per line, would you mind estimating the actual overhead in time?
>> (This is a rethorical question, sorry). I think this is microoptimization.
>>
>
> The addition of this patch is cruft to support a tool that shouldn't
> be dependent upon line-buffering at all. There are far cleaner ways
> to build applications on top of Subversion that don't require us to
> put in ugly hacks in our code.
>
Forcing line buffering would of course be silly, but it's not what the
suggested patch does. The point is to flush when there's a good chance
that there will be a pause in the output. It makes the behavior when
piping the output into a pager much nicer. It's not uncommon for command
line applications to behave in the suggested way. As was mentioned
earlier, CVS does it, and so does make. As Peter demonstrated, the patch
has no measurable impact on performance, and I don't see how it could
ever become a maintenance burden, so I'm much in favor of the patch
being applied.
/Tobias
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Feb 11 23:57:21 2006