Morning,
I been trying to trace and explain some data changes for a customer in their over night procedures, unlike Jorge im not usually awake at 3am to watch and put traces in a dozen routines so I decided to try to be clever and use our hook subroutine (MX_FILEHOOK) and output to an ascii/log of new and old data when a certain data files record is changed, then I can get up in the morning and casually take a look at the log over the morning grog.
Fantastic I thought, works so well with one small SBX change that logs every program that updates the data file records, This works fine about 95% time but I have noticed a couple of programs that do not pass the pre'rec (old / before overwritten).
So after going around in circles for while and comparing programs that work (have pre-rec populated) and those that dont I think I came to the following conclusion:
and its very old code that works a bit like this where its reading a record, then header record 1 and then writing back the original record:
! Example - KWOMORD is the record number and FWOMORD is the channel
!
! Read record
KWOMORD=25 : READ #FWOMORD,WOMORD ! Record 25 is just a random example)
! do some stuff..
! Then temporary store record number in the middle of changing.
SAV'PTR=KWOMORD
! Read file header record
KWOMORD=1
READ #FWOMORD,WOMORD'HEADER
WOMORD'REQ'NO=LST'WOMORD'REQ'NO USING "#ZZZZZ"
! Write file header record
WRITE #FWOMORD,WOMORD'HEADER
! Set record pointer KWOMORD back to original record
KWOMORD=SAV'PTR
! Write back changes
WRITE #FWOMORD,WOMORD
I hope that makes sense?
Enabling hooks code is:
hookop = HOOKOP_ADD ! add a new hook
!hookproc$ = "SBX:MFHOOK" ! Hook is handled by MFHOOK.SBX
hookevents = HFE_POST_WRITE or HFE_POST_WRITEL
hook'id = XX
hookflags = HFF_PROG or HFF_SBX or HFF_RECNO or HFF_TIME or HFF_DATE &
or HFF_USER or HFF_DATA or HFF_DATA_WAS or HFF_DATA_CHG_ONLY
XCALL MIAMEX, MX_FILEHOOK,hookop,status,hookdatafile,hook'id, &
hookproc$,hookevents,hookflags
and I just use the pre and pre'rec in the calling MFHOOK.SBX thats execute on program writes etc.
xgetargs status,hookenv,rec,pre'rec
Im testing and seeing this under my 7.0.1752.0/64 (Not tried it on Customer yet who has 6.5.1643.0)
I guess I can kind see why it no longer knows the pre-rec unless the READ for the record precedes the WRITE?
Anyway, so far as the logs is just for me I think we can work around the data i need but good to have confirmation that my conclusion is in the right area...
