I dont really class myself as a BDE hater but am definately not all that
keen on ISAMs
Under Linux I cannot see any point in ISAM support so the client will be
thinner
The difference between 'Borland' and M$ re the method of interaction
with an SQL server dont seem to be that far apart. I know MS favour ADO which
seems to be more in tune with live updates (and the devepments in MS SQL 2000
mirror this ie views with triggers and triggers that can be 'instead of' ie
define a complex view and you can insert into it as though it was a table the
'instead of trigger' updating the underlying data). This fits well with a 2
tier high bandwidth model and the table approach sits well with Access/VB/ISAM
programmers who get confused past a 1 tier system...I digress. Borland seem to
prefer 'Client Datasets' which fit well with a 2 tier low bandwidth and
'Object Models' but are these approaches so different.? IMHO it only
seems to affect the number of edited records that are held at the client ie
Live Update = 1, Client Datasets >= 1 and the manner of the posting
I wouldn't be surprised if ADO developed a delayed update method and then
the difference would disapear
PS I have not done any work with client dataset so I may be over
simplifying
Regards
Neven
|
