On 6 May 2002 [EMAIL PROTECTED] wrote: > - Remove r->boundary since it is possible to have this self-contained in > boundary's ctx. (May require MMN bump?)
That would definitely require an MMN bump. > + ctx->boundary = apr_psprintf(r->pool, "%qx%lx", > + r->request_time, (long) getpid()); Is %qx portable? Shouldn't that be APR_TIME_T_FMT? > - if (ctx->bb) { > + if (!APR_BRIGADE_EMPTY(ctx->bb)) { > APR_BRIGADE_CONCAT(ctx->bb, bb); > bb = ctx->bb; > - ctx->bb = NULL; /* ### strictly necessary? call brigade_destroy? */ > } > + apr_brigade_destroy(ctx->bb); That apr_brigade_destroy() call is unnecessary. ctx->bb is allocated from a pool, and you know it's empty because you just emptied it. There's nothing to destroy. Ie, the answer to the ### comment was "no, and no." :) PS: If you're looking for a good way to test this stuff, run a big PDF through it and view with Acrobat Reader. If Acrobat locks up, the byterange filter is broken. ;) --Cliff -------------------------------------------------------------- Cliff Woolley [EMAIL PROTECTED] Charlottesville, VA