Walter Meester
HoogkarspelPays-Bas
Hi Dragan,
>So... it may be optimized against Fox, actually. There's a big difference between "they need to clear all readbuffers" and "they need to reload these blocks which were written somewhere", where each workstation would decide for itself whether it was having anything dirty or not. The difference would show if a dozen workstations had only the initial blocks of a file, and then another workstation would update a record near the end of file. Since Fox doesn't load whole tables, just the needed blocks, reloading all buffers would be unnecessary - but if OS tells the WS to do so, it would be done regardless.
>So there's a difference between refreshing all the buffers for a table on all workstations, and refreshing only those which are out of sync. I'm rather outdated when it comes to the way today's Windows do these things, but I wouldn't be surprised if the overkill approach was chosen.
All I do know for sure is that the server is going to notify the workstation that it is breaking up to oplock mode 2, which means that the the OS of the workstation should not hold any readbuffers. However we know that VFP keeps read buffers anyways and that they´re refreshed accordingly to the SET REFRESH settings. I know that this does not mean that an actual refresh of the buffers for each file is taking place at regular intervals but more that when VFP wants to access a(nother) record, VFP is going to check if the SET REFRESH settings have been elapsed and the VFP buffer needs to be refreshed.
Again, I don´t know each and every detail, but I do have the impression there is not much more to optimize than the default mechanism between server, client and VFP.
Walter,
Précédent
Répondre
Voir le fil de ce thread
Voir le fil de ce thread à partir de ce message seulement
Voir tous les messages de ce thread
Voir tous les messages de ce thread à partir de ce message seulement