[Firebird-devel] EPP files and different ODSs

[Adriano d. S. F. <adr...@uo...>]

Hi!

As you know, to handle different ODSs we write more or less identical 
GDML code or we split sections for use  new fields in different blocks.

Both things are very ugly, BTW.

I'm doing a small patch so that we could write one GDML block using new 
fields not present in a determined ODS.

That means, we'll can clean these places in EPP files not used in utilities.

For files used in the utilities, it means we can (as new ODS appears) 
write only one block for these new ODSs and maintain the legacy code for 
older ODSs.

Patch is simple.

In PAR, when compiling a blr_fid, if the field is from a system table 
but the field is not present in the table, we change the node to 
nod_null (or a predefined/configured constant depending on table/field; 
for assignments it will be always nod_null).

That means we can read non-existent fields and write to it (assignments 
to nod_null is nop).

I think this change (only the patch) should be done in HEAD so we can 
start better handling of ODS differences in the transition branch.

Comments?


Adriano