Hi Clif,
Don't forget to read each releases patchlist in the UV directory. They have been known to put things in there that aren't in the New Features doc. The CONCURRENT option to BUILD.INDEX comes to mind.
I queried why CONCURRENT wasn't in the manuals despite pre-release publicity. The response I got from IBM support in the UK was that it is not officially released as it doesn't work. Maybe JayJay will have an update on this.
I was particularly interested to see if they had fixed a nasty bug in indexing that makes CONCURRENT useless if it's not fixed.....
UV builds an in-memory list of indices when you open a file. This is scanned whenever the file is updated to see if there are index changes to be done. If a process has the file open when another user creates and builds an index on it, the first process knows nothing about the index and and writes will immediately mean that the index is wrong.
This has been a well known bug for a long time and is one that we highlight in the UV internals course. It means that, despite BUILD.INDEX taking a lock on the file, it only makes sense to build an index when no one else has the file open.
There do seem to be other things that are in the product but not in the documentation. The new command stack editor is a good example. I can remember being at a UK user group meeting about fifteen years ago where the users were demanding this as an improvement on the "dot commands". At long last it is there but totally undocumented (unless it is hidden somewhere I haven't looked). I had also really hoped that it might recognise terminal type dependent cursor keys rather than needing control keys but it seems not. Perhaps there will be a way to do this when they document it.
The much publicised case insensitivity features also appear to be incomplete in documentation, however, my own experiments with them show that they are not really case insensitive at all but merely translate things on command lines to uppercase. I tried a simple experiment with UniVerse's own demonstration database and tried to find entries in the CUSTOMERS file that began "Mr". If I leave case insenstivity off, this works so long as I get the casing right. Of course, the whole purpose of case insensitivity is that I don't know how the data appears in the database. If I turn case insensitivity on, it never finds the records I want because it takes what I type ("Mr" in this example), converts it to uppercase, and then fails to match the data as stored in the file. I have submitted this to IBM support and the most recent response I had was that it had been forwarded to the development team.
I have also commented several times about the documentation omitting one of the most useful features of pattern matching - the ability to specify a character count range such as "1-3N". So far, nothing seems to have happened.
As a company, we are strong supporters of the U2 products and provide consultancy and training services to U2 users. Our experiences with 10.2 have not been good as we are frequently faced with comments from clients about things that don't work or are not adequately documented.
------- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
