Re: C compiler limits

In article <1p76t5INN3bi@emory.mathcs.emory.edu>, jbaker@bluegrass-emh1.army.mil ("John Baker, Blue Grass Army Depot") writes: |> Dave Kosenko of Informix Software, Inc. <davek@newjersey.informix.com> wrote |> in response to my posting about the CREATE TABLE statement and C compiler |> limits: |> |> > Limits on c structures have nothing to do with limits to the number of |> > columns in an Informix database. We DO NOT use C structures. A "record" |> > is a fixed number of bytes. We then apply offsets to break that record |> > into individual columns. At no point is this ever turned into a C |> > structure... |> |> When you say "We DO NOT use C structures," does that also apply to all the |> other SQL statements and not just CREATE TABLE ? |> |> In other words, are there ever any structures in the C code that is |> generated from the statements INSERT, DELETE, UPDATE, ALTER TABLE, |> CREATE VIEW, RENAME COLUMN, etc ? When you are passing host variables along with these, an array of Informix-defined structures will be passed (when the 4gl is translated down to C). The number of elements in the array is equal to the number of host variables involved. Since what ultimately gets passed is a pointer to the array this should not be a problem. |> Is using the DEFINE RECORD in 4GL the *only* time I have to concern myself |> with the compiler limit to the number of structure members? I am unable to commit to an answer on this, as I am by no means a 4gl "guru". Offhand, though, I can't think of anything else in 4gl that would implicitly create a structure with a large number of members. Note that my original assertion about the use of structures applied to the server end of things, not the 4gl or esql/c client end. Dave