Re: Patch: Make CGI handler obey rack spec for the greater good

[James Tucker <jftucker@gmail.com>]

On 12 Jun 2010, at 15:46, Max Cantor wrote:

> On May 20, 5:10 pm, James Tucker <jftuc...@gmail.com> wrote:
>> On 20 May 2010, at 14:36, Max Cantor wrote:
>>> Do you feel like there's a good reason to not make the CGI handler's
>>> input stream rewindable?  Every time I update rack, I have to re-patch
>>> it myself, because it never works.  It seems pretty unintuitive to
>>> require adding middleware just to make the CGI handler work at all.
>> 
>> I just don't see really good reasons to buffer for the sake of buffering.
> 
> I'm not suggesting that we buffer for the sake of buffering... the CGI
> handler literally does not work for me without this patch.  I don't
> think I'm doing anything very idiosyncratic: Just a basic rack app
> with lighttpd.  If there's some way I can reconfigure my environment
> to obviate the need for this patch, I can do that, but it seems like
> the handler's lack of adherence to the spec is making it break.

The spec went the way of requiring rewind. I'm not sure what the conditions are that $stdin isn't rewindable, but your patch is valid to solve the problem. Sadly, it just missed release :-/