D
dbeavon
Guest
We are on OE 11.5 and do a certain amount of OO in ABL. Very recently (about a year ago) we finally started noticing some enhancements being made in PDSOE to support the ability to "find references". This is invaluable when it comes to OO code (classes, properties, methods, etc). Any good IDE for an OO language will have this. Otherwise OO code can become unmaintainable (ie. like legacy ABL without any xrefs). However, a number of things are frustrating about the "find references" feature, the most important one being that it is extremely hard to believe the results. They have extensive accuracy issues. I am trying to search for a list of "gotchas" *(accuracy-related bugs) but haven't found a good list yet. Is anyone keeping a list? A list would help us know when to beware of the results (and double-check them with text-based file searching, the "old fashioned" way). Here is a sample google search: https://www.google.com/search?q=find+references+site%3Aknowledgebase.progress.com Note that near the top of the google results we see one of the biggest problems with "find references" - it takes forever to generate the results (aka. "indexing"). The generation of results is done on a background thread of PDSOE, and there is no progress/status output view for it - so you don't really know when its running, or when it's complete. On top of that I've noticed a number of other issues, even after the "indexing" is supposedly complete. PRIVATE methods are never found, not even locally within the same class, constructors aren't found in some cases (depending on the NEW A.B.Xyz() syntax where the classes are instantiated), and so on. It is really scary to rely on this new feature when making changes to widely-used OO classes. Please let me know if it is reasonable to ask for the list of "gotchas" (aka. "known issues") with the new feature of PDSOE. I have to believe there is an internal list that is kept by Progress, which is quite a bit more detailed than what we can find in the KB. As things stand now, I would almost never trust the results from this feature without double-checking.
Continue reading...
Continue reading...