In article <4e1f601fc7Jeremy@xxxxxxxxxxxxxxxx>, Jeremy C B Nicoll <Jeremy@xxxxxxxxxxxxxxxx> wrote: [snip] > What I'm trying to say is that if a server for example sent > <CR><LF><CR><CR><LF> > (or someothing like it) it's anyone's guess No, it isn't... ;-) > whether that would be seen as two or three line ends. As you can tell by looking at the second section of PROCincoming, the first <CR> marks the end of the first line. After processing it, AntiSpam will skip the following <LF> and <CR> and start reading the second line. This will be an empty one, because the first thing encountered is the third <CR>. After processing that, the program will find a single <LF> to skip and continue to read the third line. > If, as seems likely, the server is sending something it would regard as > well-formed (because a pure download of the message gets processed > properly), but AS test logic sees differently, you need to see the raw > data that the server sends, rather than what AS logs. In the early stages of development of version 1.50 I used an extra save routine to 'dump' the buffer immediately after reading from the socket. It wouldn't be too difficult to reinstate that (with some modifications to enable 'prolonged use'). Richard could use that until the problem occurred again and we could examine the saved dumps. Saving all those dumps could take up quite a lot of disk space, though. Is that OK with you, Richard? Regards, Frank