Hi,
I started using this approach back when the CF was in Beta, it works fine
from a code point of view, you can serialize a dataset with a couple of
lines of code. the only drawback was the speed, at least back them was
PAINFUL , the application just to take from 2 to 3 minute reading the
dataset, it was not acceptable.
Then I changed to CVS files, it was a big improvement but I still had to
deal with values that had comma inside, so I had to parse and interprete the
line, I used opennetCF code for this.
Finally the solution I'm using now are files files with control chars as
separators, I'm using char 16 for this, now it work much faster and I have
no need to parse the string.
I also started using XML files again as I was forced to it, in the used
platform ( iPaq 2210 ) it runs fine now, true is that this XML file is much
smaller ( 20 Kb ) than the "data" file ( 600-800 Kb ) so maybe a combination
of improved code ( CF SP 2 ) with much more powerful hardware makes it now
feasible to use XML files as data storage.
In anyway I think that you should test it and see how it works in your
environment.
Cheers,
