I guess this depends very much on the server, both on the OS and the amount of RAM. My servers are mostly WIN2000 with 1GB of RAM, so most of the necesary info is cached.
>The gain was not significant in the opening, but later, when retrieving records from the table on a file server - there was no need to check for locks on the file, and that's what was faster.
>
>>I really can not tell. My experience is that the opening of the tables is very fast, so if there is a gain, I probably would not have noticed it.
>>
>>>>Hi Fabio
>>>>
>>>>I have never had any problems with understanding what is meant: Whenever you open a table AGAIN, you will inherit the attributes, so if you must close ALL instances of the table to be able to reopen the table in another mode.
>>>>
>>>>For me this has never been a problem, I always open the tables RW except in very, very rare cases. AFAIK there is no speed gain in opening a table RO, so even for lookups I open the tables RW.
>>>
>>>Some years ago (mmmm... probably around DOS 5.0 :) it was worthwhile to open tables readonly, because it sped up the network access significantly. The network would know that these file handles are not a source of locking, so a lot of checks were dropped and it did go visibly faster. Helped with the reports a lot.
>>>
>>>So, I'd like an update - is it still so, or are today's networks smart enough with locking to the point where opening tables readonly doesn't matter much?