HFS device always busy

[Jon Tai]

I originally reported this bug in our internal bug tracker in late 2007. The HFS device was unusable because it was always busy:

GTM>D P^DI

VA FileMan 22MSC

Identity = man,sys MANAGER,SYSTEM CC SYSTEM MANAGER

Select OPTION: 2 PRINT FILE ENTRIES

Output from what File: REMOTE PROCEDURE// NEW PERSON
                                          (47 entries)
Sort by: NAME//
Start with NAME: FIRST// MAN,SYS
Go to NAME: LAST// MAN,SYS
  Within NAME, Sort by:
First Print FIELD: ALL
  Do you mean ALL the fields in the file? NO// YES
Heading (S/C): NEW PERSON List//
START at PAGE: 1//
DEVICE: HFS Host File Server
HOST FILE NAME: /tmp///file

     [DEVICE IS BUSY]

--

At the time, we hotfixed the issue by essentially forcing S %ZISOPAR=$$MODE(%ZISOPAR) to be run in %ZISF. This hotfix has been in all of our OpenVista Server releases.

The KIDS build from the first alpha release of the OpenVista/GT.M integration project does not include the hotfix, and if ^ZTMGRSET is run, %ZISF is overwritten with a version that does not contain the hotfix, so the HFS device always appears to be busy again.

When this was re-reported internally, I looked at the hotfix with the aim of including it in our build, but after more careful examination, I think that the code from the VA is fine as written - the hotfix is not required if we simply configure the HFS device with the OPEN PARAMETERS in GT.M format, i.e., "(NEWVERSION:NOREADONLY:VARIABLE)" rather than "WNS". That way, an extra call to $$MODE() is not required to translate from "WNS" to "(NEWVERSION:NOREADONLY:VARIABLE)".

[Jon Tai (jontai)]

After further discussion with Joel, we decided to include a modified version of the hotfix that will accept both GT.M-style and Cache-style OPEN PARAMETERS.