Re: found a bug in 3.13.9 dbLoadTemplate - after all that time

[kuner <kuner@mail.bessy.de>]

Marty Kraimer wrote:

> First I want to make sure I understand what you are saying.
>
> If a substitution file contains
>
> file test.template {
> { NAME="X", HOPR="9", LOPR="1"}
> }
> file test1.template {
> { NAME="Y" }
> }
>
> You are saying that any appearences of $(HOPR) or $(LOPR) in 
> test1.template should NOT be expanded.

Yes, because the definition of a value in the first template probably 
hasn't to do anything with the value of the second one but the name. 
Every template has to have its own macro expansions. The behavior now 
avoids an error when the .db will be loaded and sets it to anything. 
Mistakes like this cause a bad behavior of the database and are not easy 
to find.

> Both dbLoadTemplate and msi (Macro Substitution and Include) expand 
> $(HOPR) and $(LOPR) using the last definition they received. Thus in 
> the above they will be expanded using HOPR="9", LOPR="1"

krautig, gruselig, bad and evil.

> I can see either semantic definition being valid. Since the existing 
> semantics have been in use for many years, it does not seem to be good 
> idea to make a change. Many users may rely on the existing semantics.

And how many of them are traped by this? You mean a .substitution file 
like this should work??

 file defineAnything.template {
   { FIRST="1",SECOND="2",THIRD="3",VIERTER="4",FUENFTER="5",SECHSTER="6" }
 }
 file takeSomething.template {
 }
 file takeSomeOtherThing.template {
 }
 file takeItOrLeaveIt.template {
 }

This will expand to... to what? (May be, C-Hardcorepragrammer would like it)

best regards

Bernahrd Kuner

--
-------------------------------------------------------------------
-- Bernhard.Kuner@bessy.de ----------- Bessy Gmbh -----------------
-- +49 30 6392 4981 ------------------ Albert Einstein Strasse 15 -
-------------------------------------- 12489 Berlin ---------------