Re: Validating a File Member Name

Was the plan to perform "efficiently" dismissed as a requirement? That CLP would perform significantly worse plus waste significantly more resources than even the RTVMBRD someone else suggested. A very poor implementation IMO, and seems completely to miss the intent of using the syntax checker versus actually invoking any commands.

Since the QCMDCHK API was apparently ignored, I do not suppose there is much benefit in suggesting another.? There is also the Verify Name (QCAVFYNM, QCAVERIFYNAME) API for which CPF019D is sent for an invalid name; use the format VFYN0100 and the *NAME for "basic name" as alluded on the example *CMD CHKNAM.

Verify "an input value to determine if it is a valid system name"
http://publib.boulder.ibm.com/infocenter/iseries/v6r1m0/topic/apis/qcaverifyname.htm

FWiW I did not suggest the above API in my first reply, because although I was convinced such an API existed, I could not recall the name, nor have I ever coded to that API; I have used QCMDCHK for that effect however. And I did not see the doc link in the InfoCenter search results in an earlier search on _validate name API_ but a new search omitting the token API was as good as when searching on the _verify name API_

Regards, Chuck

On 09-Sep-2011 12:41 , Vinay Gavankar wrote:

In my environment, I won't be allowed to "create" a new command.

But using the idea I wrote a small program:

<<SNIP "ugly" CLP source>>

Assuming @@WRK@@ in QTEMP is not a file I will have normally, this
seems to work.

Any comments?

On Fri, Sep 9, 2011 at 10:09 AM, CRPence wrote:

On 09-Sep-2011 06:55 , Vinay Gavankar wrote:

What would be the most efficient way of validating that an input
provided on Screen is a value that can be used to add as a Physical
File Member later in the job stream (i.e. it starts with alphabet or
specific special characters, no embedded blanks, etc.)

I feel that it should be possible using some system commands and
MONSG etc. instead of writing RPG code to check, but cannot figure
out what it should be.

Very possible. The Check Command Syntax (QCMDCHK) API suffices.

The most obvious though perhaps not generally the most efficient
would be to use the command that will eventually be utilized. For
example, if the value will be used in ADDPFM, then generate and
pass a valid command string with the MBR() parm value concatenated.
For an idea of how, some command-line examples:

/* .........+.........+.........+.........+.. */
call qcmdchk ('*nlvlibl/addpfm literal/literal tenbytenam' 42)
call qcmdchk ('*nlvlibl/addpfm literal/literal 9bbadname ' 42)
Value '9BBADNAME ' for parameter MBR not a valid name.
Errors occurred in command.

Somewhat more efficient since the command is minimalist... Using
the same means above but with a simpler user-created command such
as CHKNAM:

//data /* mbr(chknam) */ endchar('//')
PARM KWD(MBR) TYPE(*NAME) LEN(10) MIN(1) MAX(1)
//
crtcmd chknam

call qcmdchk ('chknam validname ' 25) /* no errors logged */
call qcmdchk ('chknam *errname ' 25)
Value '*ERRNAME ' for parameter MBR not a valid name.
Errors occurred in command.

In the error cases, validation for CPF0006 as the *ESCAPE and then
thee preceding diagnostic being CPD0078 giving the invalid value
in replacement text variable &3. When no errors, the value in the
string can be assumed to be correct when concatenated to the
eventual complete ADDPFM command string.