Update data dictionary on Network

[Tom Patten]

Hi,

When on a network, using network optimization, when referring to tables to update data dictionaries, does one reference the table through the local shadow database, or does one reference the actual data table (X:drive)on the network?

Or does network optimization update dictionaries making an additional update unnecessary?

Thanks

Tom

[Jim Chapman]

RE: Update data dictionary on Network

Hi Tom,

Network optimization creates a new, or overwrites and existing, optimized database. This optimized database contains all the data dictionary files and a table file that 'points' to the real table file. A shadow_refresh copies the data dictionary files but does not update the shadow tables. If you don't change the table structure a refresh should be all you need. That said, I've had problems with the shadow refresh. Sometimes it appears to work but then gives an error when you try to open the shadow. It usually give an error message that would lead you to believe there is a path problem. Other times it just doesn't appear to work. I always do a complete network optimize whenever things need updated. Unfortunately, xbasic doesn't provide in version 4.5 the tools to properly automate this process. Sys_send_keys() can be used, but there are problems with this also. If I automate a network optimization in v4.5 from an existing shadow, it almost always works the first time, almost always doesn't work the second time. I think it has to do with alpha's name space arrangement. Good news is that v5 gives us the tools.

Jim

[Tom Patten]

RE: Update data dictionary on Network

Hi Jim,

Thanks for responding. What I find is that occasionally I need to update indexes manually when on one computer. If I put my application on a network, if I want to update indexes in a table, do I address it thru its shadow, or address the network drive, in this case X:?

Or would I network optimize again to force tables to update indexes?

Tom

[Allen Klimeck]

RE: Update data dictionary on Network

This will update your indexes but it does no error checking. No other users can be accessing the table when you update. Just put it on a button on your main form not on the open table.

tbl = table.open("Your_table_name")
tbl.update_production_index()
tbl.close()

[Tom Patten]

RE: Update data dictionary on Network

Thanks Allen,

I have a script that goes through all my tables and updates indexes, what I want to know is, when installed on a network, how does one reference the tables;

e.g. tbl = table.open("c:|A5V4|shadow|etc...) 'LOCAL DRIVE
OR tble = table.open("X:table reference) 'NETWORK DRIVE

I'm guessing the latter, but...

Tom

[Jim Chapman]

RE: Update data dictionary on Network

Hi Tom,

You'd reference your local shadow table just like normal, A5 takes care of the rest. You could address the table directly on the server, but I don't see a reason to do this in normal circumstances. Alpha will give you the path to the 'real' table via the 'a5.get_master_path()' function, but again just reference the table as you normally would. Indexes of course have to reside on the server as everyone on the network has to have the same indexes. Queries however reside locally.

Jim

[Allen Klimeck]

RE: Update data dictionary on Network

If all of the tables are in the same folder you don't need the path at all only the table name server or workstation.

[Tom Patten]

RE: Update data dictionary on Network

Thanks guys - you stay up later than I do!

Tom

[CALocklin]

RE: Update data dictionary on Network

Sorry guys, I don't totally agree with this.

First,a disclaimer: I'm not sure what "update_production_indexes" does because I seldom use it. I use an Xbasic script to completely rebuild the indexes because I want to make sure all extraneous indexes have been removed and, more important, all required indexes still exist. Unfortunately, I've run into situations where users delete indexes because they think they aren't necessary.

If you want to update the indexes, I recommend accessing the network file itself. I know that running the following script in a shadowed application will update indexes only on the local computer and not the network:

tp = table.open( "MyTable" )
table.index_create_begin( "LastName", "Last_name" )
tp.index_create_end()

To fix this you need:
tp = table.open( a5.get_master_path() +chr(92)+ "MyTable" )

If you want to find out what's happening on your system, try this:
1. Delete any .cdx files in your shadow directory - they shouldn't be there.
2. Run your index update.
3. Check your shadow directory again for .cdx files. If any are found, you didn't update the indexes on the network.

[Jim Chapman]

RE: Update data dictionary on Network

Hello,

The table index file resides on the 'server'. When a shadow is created it does not bring in (copy over) the *.cdx to the shadow. The tables in the shadow are just a pointer to the real tables. This has to be, as there can only be one copy of the real data, just as there can be only one copy of the *.cdx file. Any index (not query) you use is the same one all users have access to.

In response to Cals' post I checked this out and he's right, .index_create_begin() does not work when run from a shadow without feeding it the master path. It will work if you point it to the server. I'm not so sure whether this is by design or if it's a 'feature'. Maybe it's a safety feature, but I doubt it. *.index_create_begin() creates a whole new production index, doing away with the old and replacing it with the new tags that you supply. *.index_tag_add() works as advertised, pointing it to the shadow table adds an index (tag) to the production index file on the server. *.update_production_index() also works by pointing it to the shadow.

The "update_production_indexes" command updates all the tags in the production index file (*.cdx), just as if you right clicked on the table icon and choose 'Update Indexes'.

Jim