i-scream documentation viewer
minutes-20010129.txt
Minutes of meeting 29/01/2001 @ 2pm Location: UKC Computer Science Meeting Room Present: ab11, ajm4, pjm2, tdb1 Absent: None Paul informed the group about the host he had written in Perl during the week. It has most of the required features of a "basic" level host, and is more stable (due to regular expressions) than the Java host. Tim reported that the host (named ihost) was running fine on all the rocks, and had been for over 12 hours. Tim reminded the group that another stab at a "24 hour" run was in progress, which had gone over the 10 hours of the first attempt. This was put down to the quota increase on raptor for mySQL. Paul reminded the group of the changes to the database. Over the week he and Tim had discussed ways of improving database access, as 2 million rows for 10 hours was ridiculous. It was decided that a "flat" table with the XML string stored in a complete state would be best (with a few key fields extracted). Not only should this reduce the number of rows, but it should reduce the disk usage. Paul reported that this had already been implemented on the server side, and data was being added using this method. AJ told the group about the progress with his client. Although there were only a few "cosmetic" changes, a lot of work had been done with the "behind the scenes" stuff to make it better implement the server protocol, and to make it more Swing thread safe. AJ then informed the group of his plans to improve the configuration side of the client, to make it fully implement the server protocol, and to improve the visulisation of data. Paul stated that the DBReporting tools would now need to be edited to use the new database format, and that he would work on this. He also said he'd get some more done on the command line client, although it was becoming tricky to make it work under both DOS and Unix. Ash reported that the host development was going well, and all the central structure was in place. The "data grabbing" part still needed to be coupled with the Perl script, and the networking module had yet to be done. Tim added that he had some good stuff on networking, and would dig it out. Ash also discussed some of the ideas behind alerting which both the host and the "local clients" would need to be based around. Tolerance levels will be stored in the central configuration, allowing them to be retrieved by both the host and the client. The host would use an "averaging" method to send data over longer periods of time, but based on these tolerances could send priority packets when something "bad" happened. Tim was last to chip in and tell the group about progress with the server. The client interface was complete for TCP clients, such as those AJ and Paul are writing. The rest of the server is complete, and bugfixes have been carried out throughout the system as they are found. Tim suggested a "feature freeze" in the server for a while to allow work to continue on the rest of the system, and to provide a more stable testing platform for hosts and clients. The group agreed this was a code idea. Tim said work was to continue on "patching and fixing" the server, but that he would now focus on setting up the architecture of the new "local clients". These would use a CORBA setup to connect into the Client Interface part of the server. Further details to follow when design has been completed. On a final note the group discussed the benefits of demonstrating the system to jc (this was later backed up by iau). The group agreed to get the client looking better, the reporting tools ready, and the local clients in a decent state before this could happen. It was noted that the feedback from a potential end user could be invaluable to the project, but that time must be given to implement any changes resulting from this feedback. There will be a meeting as usual with Ian, but at the later time of 2.15pm (now a regular time). The next group meeting will be next Monday.
