Hi Tore,

>For more than 25 years I have NEVER needed to contact any supplier for technical support, except for a few cases where the documentation was either missing, inadequate or plain wrong. So I really don't understand all this fuzz. Or is it only me?

Dunno... I consider myself to be able to work around issues, but there are cases were the work-arounds can become a major obstacle: I ran into this when I encountered a bug in the use of indexes based on vfp-functions. Aleksey gave me a fast fix oodles better than my workaround and fixed the bug for SP2. I am sure being able to trace vfp's sources made finding the fix easier (I don't think I'ld have tried the direction, even if in hindsight it seems rather logical <g>).

Another issue: secure communications (SFTP, FTPS, local certificates, web services). Here I strive to use finished components, and with vfp I encounter somtimes COM problems not found in other languages. Writing wrappers just for vfp COM is not a very sound strategy as it increases the fragility. I could reolve the issue in one component where vfp did not like the COM-variant array parameters by creating another routine working directly off strings - but other vednor's willingness to create such modifications for vfp surely has dropped even lower.

So while I am not one of crowd chanting "the sky is falling", I expect more troubles ahead - for instance currently I am investigating perhaps another bug only apparent in machines > 512MB and seemingly unrelated to sys(3050) settings, Swap file size and occurring after heavy processing only. As I expect data loads to increase and my machines all have > 1.5 GB of RAM, this does worry me A LOT - processing takes 3 times as long on artificially crippled machines. If large physical RAM creates problems under W32, W64 is not guaranteed to give you more options for intelligently using more RAM, and I will have >4GB machines later this year. Having a disk cache of 3-5 GB in 64-Bit could help me - but if vfp barfs globally in such surroundings, even if using less than 256MB for itself via sys(3050), I will be hard pressed to argue for continued use. As much as I personally like the fox. Windows memory mangament is an arcane art of its own, and I doubt I will try to watch/fix working set size or other memory parameters. And yes, I try to accelerate testing in such surroundings, but this is very time and resource consuming - even if I bill only a small part of such efforts.

regards

thomas