JACAL Release Stuff: - Develop script to perform/assist installation - Get rid of, or give credit for all proprietary pieces. - Change to more unified documentation system being used in other places - Test Windows 2000, see how well unattended and app installs work Internal Functionality: - Add option to filelink.pl to take app. dirs as an option. This would force a relink only of those application conflicts. - Change zipinst.pl to actually process the INF to set which location contains app share. - Add ability to zipinst.pl to pull packages from different locations (locally, server, etc). - Add ability to DLL resolve script to resolve across multiple locations and servers. (going along with above item). - Add ability to install net based apps locally, (i.e. pull down shared components) - zipinst.pl should be able to install from package list. This list should have mode (local, or network based) - Create zipremove.pl to remove an app. It should just have to pull the stuff from the install source directory and ignore shared info, hopefully that should work.) - Get DLL resolve script to detect conflicts in registry entries as well. Very Long Term Goals - Replace the sysdiff utility with something that does things properly. For example, creates an INF file of changed registry entries and not just added entries. Also, something that goes faster. Perhaps have it detect when apps read registry entries so that it can make guesses about which entries are linked to other entries. Possibly reverse engineer DISCOZ.EXE (Discover) file format. - Use NT _ONLY_ to perform the NT installation step. When Linux NTFS write gets good enough we should do app installs and registry updates from Linux (i.e. after NT is installed and reboots.) - Local JACAL. Have JACAL and the initial NT install files stored on a local partition. Perhaps use a boot manager which detects whether the machine is scheduled for a rebuild. The rebuilds, or otherwise simply reboots. Could be used to control number of reboots for NT install etc. - Create FCP. File Cast Protocol. Get our circular file multicast protocol working so that we can speed up large file transfer dramatically. This would make file transfers linear according to the amount of data to be transfered rather than the number of machines.