Thanks Phil for the idea, this could indeed work.
[ Been using something similar for real .dbfs.
It was quite though to implement properly, but luckily
for mem files some of the problem cases can be disregarded. ]
Brgds,
Viktor
On 2009 Oct 5, at 08:33, Phil Barnett wrote:
On Tue, 2009-09-22 a
On Tue, 2009-09-22 at 18:23 +0200, Viktor Szakáts wrote:
> Yes, with real FS it's okay with our APIs, but
> such date/time/random based solution aren't rock
> solid (uniqueness isn't guaranteed), so indeed
> something better would be good.
pseudocode
do
RandomFileName = SomeRandomFileNameCrea
Viktor Szakáts wrote:
My very first problem towards implementing hbmemio
as a replacement for temp tables is that I'd need
to create *unique* "filenames".
BTW the same problem goes to hbnetio, where I sometimes
need to create unique temp files (which later get renamed
to final table name)
How to
Hi,
Viktor Szakáts wrote:
My very first problem towards implementing hbmemio
as a replacement for temp tables is that I'd need
to create *unique* "filenames".
BTW the same problem goes to hbnetio, where I sometimes
need to create unique temp files (which later get renamed
to final table name)
