Re: Processes after CRTSRVPGM

["M. Lazarus" <mlazarus@xxxxxxxx>]

Scott,

 A software vendor installs a *SRVPGM to be utilized by clients' software.

Scenario:
 Signature is hardcoded.

 Version 1 has two exports.

 Version 2 with 3 exports gets shipped and installed.


 If the manual signature has not changed, the new version will not 
throw an error if the calling programs have not been rebound.  This 
may be a desired behavior, if the new version handles it 
properly.  But the oversight may allow undesirable results if an 
export was dropped.

 -mark


At 2/14/08 12:27 PM, you wrote:
> Mark,
>
> The signature on a service program does NOT work like a level check on a
> file.  Using *CURRENT/*PRV with a generated signature does NOT detect
> ANY additional errors over using a hard-coded signature.
>
> If you're using a *CURRENT block *WITHOUT* using a *PRV block (but still
> using a generated signature) then it does offer some protection, but
> throws away one of the biggest values of ILE -- because now every time
> you change the procedures that are exported from the service program,
> you have to recompile everything.
>
> There's no value to using *CURRENT/*PRV.  If you disagree, please
> outline a situation where it'd protect you, but a hard-coded signature
> wouldn't.
>
>
>
> M. Lazarus wrote:
> > Buck,
> >
> >   If you would have the ability to directly set the level ID on a
> > file, would you?  In the vast majority of cases, of course
> > not!  Because the system knows to generate a new level ID / signature
> > *automatically* when a significant attribute changes.
> >
> >   Why wouldn't you want similar protection for binder signatures?  I
> > think that signatures fall short as it is (no sig change if the
> > attribute changes, etc.), why would you want to short change it even more?
> >
> >   -mark
> >
> >
> > At 2/13/08 03:54 PM, you wrote:
> >> Adam Glauser wrote:
> >>> M. Lazarus wrote:
> >>>> Lim,
> >>>>
> >>>> [Specifying SIGNATURE on STRPGMEXP] tells the system that you are
> >>>  >  managing the signatures.
> >>>> This is almost the equivalent of setting LVLCHK(*NO) on your files.
> >>>> It gets rid of the error message, but also disables the error checking!
> >>> That's a bit of a stretch in my opinion.  It's true that specifying a
> >>> signature can cause problems.  However, I think it's quite a bit simpler
> >>> to recreate or update a service program safely while overriding the
> >>> safety measures than it is to add a field to a file with LVLCHK(*NO)
> >>> without causing problems.
> >> I agree with Adam.  I think the difference is that with files, there is
> >> no *PRV support, whereas with binder language, we can successfully use a
> >> newly compiled service program with an old calling program.
> >>
> >> I have a strong tendency to make my signatures look like strpgmexp
> >> pgmlvl(*current) signature('01.02 20071019').  When I do a DSPPGM I can
> >> immediately see what version of a service program it's bound to.
> >>
> >> The caveat, of course, is that one must add new exports AFTER existing
> >> exports or the calling program will call the wrong procedure without a
> >> signature violation.
> >>    --buck