Providing support for all E. Crane Computing products, including POWERGEN, PBSERVICE, HARPB, VERSIONEDIT, PBLRESCUE, and PBLEXPLORER.
You are not logged in.
Pages: 1
Our current app uses PB8 and we deploy in p-code. I am researching ground rules for my developers and builder as to what object changes force a complete (full) regen and build of the application we are maintaining. So far I have found nothing informative on this topic at the various web sites so I thought I would come right to the source with my question. We have been using PowerGen for several years now and the need know if a set of object changes will require a full build has become important with our client lately. Generally we have found that a full regen/build is needed when we have changes in the following areas.
1. Menus – Adding or removing an item will cause the need for a full build.
2. Any change to the nv_transaction object can cause the need for a full build.
3. Any change to ancestor objects can cause the need for a full build.
4. Any change to objects in the core pbls (pbls found in the core folder) will cause the need for a full build
5. Any change to objects in the common pbl will cause the need for a full build
Do you have any further insight as to what would cause the need for a full regen/build? We release full builds six times a year but are also trying to release partial builds as needed. Thank you for your help.
I'm not sure I'm "the source", but I'll offer my view.
An object needs to be regenerated when it contains a reference to any other object whose "signature" has changed or to a changed object that is an ancestor. By signature I mean global variables, public function definitions, etc.; those elements of an object which are "visible" to other objects.
Determining this is inherently a recursive process, since if, for example, Function A needs to be regenerated because it references a changed object, then you need to look for objects that reference Function A for regeneration. And so on.
In the latest download of PowerGen (V5.5.1) the analysis done for incremental regeneration is as I've described it above. In other words, it's as rigorous as we know how to do it. You shouldn't use it blindly (i.e. test, test and test), but I'm a believer that if the incremental regeneration process is robust, you should be able to distribute a subset of the PBDs.
Let me know how this works for you.
Regards,
Phil
Pages: 1