Using Arev:

What is the problem/fix when a "Select Table_name" returns 48 record keys and a "list table_name" returns 53. I have added new records to the table and they only show up in the "list"…not in the "select."

Also can't delete records from this table.

I have rebuild indexes…no help.

Any suugestions?

Thanks,

Chris

If you DUMPLH tablename at TCL how many records are reported - similarly with a COUNTROWS tablename?

The Sprezzatura Group

World Leaders in all things RevSoft

Thanks for quick response.

Both report 53 rows

Chris

OK it seems likely that you have 53 rows then. What does SELECT TABLENAME WITH @ID ] '' report?

The Sprezzatura Group

World Leaders in all things RevSoft

Great. You give me enough clues that I found the problem. Somehow a couple columns in the key were blank. Therefore a 'select by' did not include those records, but a 'select' and 'list' did.

Now, my problem is how to delete a row with a bad key.

Thanks

See here for a non-Sprezz response

http://www.revelation.com/__8525652b0066bfaf.nsf/0/8E999C9BD28D166085256B05006C9E34?OpenDocument

The Sprezzatura Group

World Leaders in all things RevSoft

If you're in AREV:

1. Remove any QUICKDEX / RIGHTDEX, B-Tree on @ID

2. SELECT

  This should give you all of the rows (even with blank keys)

3. EDIT

When you get to the rows with null keys, ALT+D. This will USUALLY delete the row.

Othwrwise, you can write a quickie (as in Sprezz's link) to find/delete the trash.

As an aside, you should check your code or form which created these records VERY carefully since it allowed a null key to slip by.

HTH

Don Miller

C3 Inc.

I've always wanted to write NONULL.MFS….even better would be INVALID_CHAR.MFS, which checks for nulls, delimiters and other customisable chars.

Maybe for the next SENL, if there's a call for it.

The Sprezzatura Group

World Leaders in all Things RevSoft

Good idea .. if you have lots of time on your hands. I have a UDC that does something similar .. returns an error if passed data is null or contains reserved delims. Probably easier than an MFS in one way since it traps the error at the source (screens & programs) with an ugly MSG box. If you'd like a copy let me know.

Don M.

pmfji

The problem with a UDC is you HAVE TO call it.

If it needs to be a primary business rule shouldn't it be in an mfs .. then it can't be bypassed

2cents

dsig@sigafoos.org onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)"

David Tod Sigafoos ~ SigSolutions

Phone: 971-570-2005

OS: Win2k sp2 (5.00.2195)

OI: 4.1

Good point.

As an aisde, we haven't done any MFS benchmarking in a while- I wonder if there is still a 10% hit on even an empty MFS…

Just thinking…

Some RBASIC "BRANDS" executes the "ELSE" clause of the "WRITE" command (WRITE fails) when the KEY (ROW ID) is NULL or has invalid characters.

Maybe a future upgrade??? And then… wait a second…

See, the file structure was designed to allow any characters in the KEY (ROW ID) and the ROW (RECORD) itself. Actually, I don't think the filing system has the problem.

Now I do know that having system delimiters in the KEY (ROW ID) can cause other items to have troubles. Namely the "SELECT" and "INDEXING" routines.

I, like Mike, solve those problems by myself. Don't let the "WRITE" statement use KEYS (ROW IDS) with invalid characters.

Stop tossing out these challenges - like we're not busy enough already

The Sprezzatura Group

World Leaders in all things RevSoft

100,000 ROWS OF 100 "*" no mfs took 44 sec

100,000 ROWS OF 100 "*" with MFS.21 took 46 sec

dsig@sigafoos.org onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)"

David Tod Sigafoos ~ SigSolutions

Phone: 971-570-2005

OS: Win2k sp2 (5.00.2195)

OI: 4.1

Way to go Dave! How about adding SI.MFS and seeing how much it slows down READs?

The Sprezzatura Group

World Leaders in all things RevSoft

You're undoubtedly right. Being lazy, I took the easy way out.

Don M.

Sorry .. travelling between pdx and denver .. first chance to test the additional requirement ..

1) Create New Table, read 48K recs from existing table and write to new table

  # of recs:     48827

  Avg Len:       300.7 

  Total Time:    171 seconds

2) Create New Table, add btree index for field 2, read 48K recs from existing table and write to new table

  # of recs:     48827

  Avg Len:       300.7 

  Total Time:    271 seconds

3) Create New Table, add btree index for field 2, place MFS shell before index (param 2 on set_mfs) read 48K recs from existing table and write to new table

  # of recs:     48827

  Avg Len:       300.7 

  Total Time:    366 seconds

dsig@sigafoos.org onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)"

David Tod Sigafoos ~ SigSolutions

Phone: 971-570-2005

OS: Win2k sp2 (5.00.2195)

OI: 4.1.2

PII 300 laptop

MFS was compiled without linemarks?

no .. just basic compile .. could get a few seconds back

View this thread on the Works forum...