Our group is involved in expanding the Flood model to test parts of a page in parallel. As Flood seems to rely more on the multithreading model than on parallel processes (the build logic will go that way as long as the support is found in the APR as built on the current operating system, and the current code runs the processes of farmer Joe only as per the recurring Flood status emails when threading is not found in the APR), it is important to know the limitations, if there are any significant ones in the operating systems which are likely to be used. As we are building our models on Windows as a primary environment, it would be good to know if anybody can think of any limits to scaling with the Windows multithreading model using Flood. In order to use our model efficiently, I will have to make changes as well to the utilization of the pools. I have looked at the APR libraries to verify this, but I just wish to verify with the developers of the library (and others "in the know" on the list) that I am not misusing them. My understanding is that creating a pool from another pool does not pose a limitation on the original pool, it just causes the second pool to inherit characteristics of the first (and also links them for the destruction of the pools). Therefore, allocating from the first pool after the second pool has been created (and also after it is used) should be a safe operation. If anybody has knowledge or experience to the contrary or desires to elucidate further, please let me / the list know.
-Norman Tuttle, Software Developer/Consultant, OpenDemand Systems [EMAIL PROTECTED]