Re: [twsocket] GBs of Memory leak which Memproof says is at the line Send(FDocBuf, len)

2010-09-10 Thread Fastream Technologies
Working like a charm now! Thanks a lot goes to Arno (as usually!). On Fri, Sep 10, 2010 at 5:42 PM, Arno Garrels wrote: > Arno Garrels wrote: > > Fastream Technologies wrote: > >> Let me elaborate: As our code is a proxy server and data keeps coming > >> down in OnDocData of the Httpclient, I w

Re: [twsocket] GBs of Memory leak which Memproof says is at the line Send(FDocBuf, len)

2010-09-10 Thread Arno Garrels
Arno Garrels wrote: > Fastream Technologies wrote: >> Let me elaborate: As our code is a proxy server and data keeps coming >> down in OnDocData of the Httpclient, I was pumping it to >> THttpConnection's descendent with Send(buf, len). However I noticed >> it was actually buffering data when the S

Re: [twsocket] GBs of Memory leak which Memproof says is at the line Send(FDocBuf, len)

2010-09-10 Thread Arno Garrels
Fastream Technologies wrote: > Let me elaborate: As our code is a proxy server and data keeps coming > down in OnDocData of the Httpclient, I was pumping it to > THttpConnection's descendent with Send(buf, len). However I noticed > it was actually buffering data when the Send()'s pace is less than

Re: [twsocket] GBs of Memory leak which Memproof says is at the line Send(FDocBuf, len);

2010-09-10 Thread Fastream Technologies
Hello again, I have found out that there are two linked lists in wsockbuf: inuse and free. And even the allocated memory in free is not freed in some cases. It must have been intended for reusing but it is no good to keep hundreds of MBs of unused buffers in memory! There must be a limit to the re

Re: [twsocket] GBs of Memory leak which Memproof says is at the line Send(FDocBuf, len);

2010-09-10 Thread Fastream Technologies
Let me elaborate: As our code is a proxy server and data keeps coming down in OnDocData of the Httpclient, I was pumping it to THttpConnection's descendent with Send(buf, len). However I noticed it was actually buffering data when the Send()'s pace is less than Receive(). So it might be a special c