On 09/27/2009 07:43 PM, Jeff Trawick wrote: > On Sun, Sep 27, 2009 at 9:33 AM, Guenter Knauf <[email protected]> wrote: . - I assume the below is the correct fix: >> >> Index: ssl_engine_io.c >> =================================================================== >> --- ssl_engine_io.c (revision 819298) >> +++ ssl_engine_io.c (working copy) >> @@ -1665,7 +1665,7 @@ >> } >> else { >> /* Split a line into the passed-in brigade. */ >> - rv = apr_brigade_split_line(bb, ctx->bb, mode, bytes); >> + rv = apr_brigade_split_line(bb, ctx->bb, block, bytes); >> > > looks reasonable here; I guess that with this function the bucket is always > in memory, so the potential mismatch of [non]blocking didn't have any > operational impact >
+1. Looks like the current code has a bug here, but for the reasons Jeff mentioned it never made any difference in practice. Regards Rüdiger
