We are about to publish software which runs over Unidata and Universe for Windows. What is the preferred deployment package for Universe? 1) Full account, source and object in Type 1 files, Pick flavor. 2) Full account, source and object in Type 1 files, Ideal flavor. 3) Zip with object and source files, instructions to create an account of preferred flavor and move code into that new account. 4) Something else?
For Unidata is there any common/preferred mechanism other than a zip of a full account? For proper cross-flavor deployment, if I use any specific options in our code, will the behavior be overridden by site-specific preferences? Of course this code won't function properly if preferences are changed out from under it. How can I guarantee that the code will function in another environment as it does in our testing systems? Does code need to be recompiled over Linux and AIX, or can I deploy the code and data using style #3 above and expect it to work over *nix? I've accommodated differences for "SH -c" vs "DOS /c" and other OS-specific nuances, but I'm not sure if the object code itself is fully portable. This untested *nix deployment will be provided as a beta and I will get the resources required to do full OS testing if required. Finally, to accommodate the intended audience, should I be targeting another OS or flavor in addition to those mentioned above? Any other considerations? Thanks! Tony Gravagno Nebula Research and Development TG@ always.remove.this.pleaseNebula-RnD.com ------- u2-users mailing list [email protected] To unsubscribe please visit http://listserver.u2ug.org/
