I was testing MAKE.INDEX in an AREV 2.1 system and followed the BTREE example in Sprezzatura RevMedia FKB V1I3A3 and got myself into a bit of a problem. I checked my program and it seems to be same as example.

The file that I was attempting to add an index to is now not available.

All attempts to access via EDIT, LISTDICT, DICT, gives the FS105 message "Physical error Reading LOCAL.GROUP in file ???"

I can only access the dict info by EDIT DOS C:\APP1\REV64002.LK (the dict of the file) and can look about but nothing stands out.

I guess that I can restore DATA and DICT from back up and try again but I want to find out what happened.

Any ideas?

Seems like a basic GFE type error. Can you run a LHVERFIFY on the file?

The Sprezzatura Group

World Leaders in all Things RevSoft

LH_VERIFY on the damaged file does not work and gives same error message FS105 - Physicl Error Reading LOCAL.GROUP.

LH_VERIFY on other files are ok

Ted

I'd try to copy the dictionary records one by one to another file or DUMP the file and try to find the bad group and fix it. As a dictionary, the file shouldn't be too large.

Alternatively, you could just try and move it to another part of the disk.

The Sprezzatura Group

World Leaders in all Things RevSoft

I can DUMP and LIST !INV (the damaged file)

I cannot DUMP or LIST INV

Is there a reference to LOCAL.GROUP in the file (not !file) ?

Since I have EDIT access to !INV if found references to the index that I just added which was CITY Btree.

What do you know about the structure of the index files? Does Sprezzatura have the detail posted somewhere?

I found in !INV the following:

record CITY*1 =2 @vm ALL.IDS

record CITY =AL @vm CITY @vm CITY

record CITY* =1

record *INDEXES = a list of indices with the @vm CITY added at end

record !INV =list of indices with @svm CITY @FM @FM 0 @FM 4 at end

(I interpret the edit display a small 1 is a @FM, small 2 is a @VM, small 3 is a @SVM)

I am tempted to remove all references to the CITY index and see what happens. What do you think?

Ted

PS - To whom am I corresponding? I feel uncomfortable just speaking to the generic "The Sprezzatura Group"

Progress update

1 - Restore dos files from backup

- I restored the dos INV, DICT.INV and !INV files from backup

- Re-attached

- Still no access to INV DICT.INV

- Access ok to !INV

- problem still same as before

2- Delete and re-create INV

- Deleted INV

- re-created INV

- re-created 1 of the indices - this made !INV

- copied dos level from backup to the new dos files with rename

- now have access to all data, all dict and all indices.

- problem solved

Conclusion:

- The MAKE.INDEX modified some datafile not in INV , DICT.INV or !INV

- modification made incorrectly.

- This is related to LOCAL.GROUP, %LOCAL.GROUP%, International Environment, Local character sets, etc. I am not sure of the logic.

Help requested:

- Do you know where the changes might be so that I can fix in the future more easily than the above?

- Is there a document that expains in detail logic of indexing down to the level that I need here?

- do you have detail logic of MAKE.INDEX ? (Source maybe?)

- does Works membership allow access to this OLD source code?

Thanks for the help

Ted

As far as I'm aware, LOCAL.GROUP is only used for QUICKDEX indexes. The error would not have been related to the anything in specific, but is a generic GFE or file access problem. It's just a coincidence that it's affecting LOCAL.GROUP.

The copy on backup must have had the same corruption.

As for identification, my designation is Three of Six.

The Sprezzatura Group

World Leaders in all Things RevSoft

Thanks 3 of 6 for the help.

I will now continue testing MAKE.INDEX to see if I can get it to hang together and do something useful rather than trashing by aps. This time I will test on copy and not on live volume.

PS. If you are an x-Borg and look like 7 of 9 then pls send pics and bio.

Ted

View this thread on the forum...