Re: DBD/DB validator?

[Andrew Johnson <anj@aps.anl.gov>]

Switching to core-talk for development discussion...

On 04/23/2015 11:55 AM, Michael Davidsaver wrote:
> The complicate is that all the parsers in Base are designed to work on
> full definitions.

Yup, and in 3.15 dbStaticLib can't load any record instances until after
the call to RegisterRecordDeviceDriver() either, so you can only use it
to write tools which must be given all the record types in advance. The
3.14 dbStaticHost library didn't have the registration requirement (see
src/dbStatic/dbStaticNoRun.c) but you still had to load the record types
before it could read any record instances.

> dbdExpand.pl from 3.15 doesn't recognize the "record" keyword.

I've just added support for record parsing to the Perl parser, and a
dbExpand.pl script that outputs just the record instances. This won't
solve the above issue though, it still requires you to load the record
type definitions before it will accept any instances of those records.
Not committed yet, I should probably push this out on a branch for
review first.

> So I can see two courses.
> 
> 1) Modify the makefiles so that the validator can be given .dbd file(s)
> as well.  This would allow (an updated) "dbdExpand ... >/dev/null" to
> serve as a check.

Ralph also wants to be able to standardize his DB files, for which
dbdExpand.pl would be no use, but my new dbExpand.pl script could help
with that.

> 2) Write a new utility which parses, but doesn't follow include
> statements, then does as much checking as possible.

It looks like several people have already done that. I don't agree with
you about the include statements though.


- Andrew
-- 
Light thinks it travels faster than anything but it is wrong.
No matter how fast light travels, it finds the darkness has
always got there first, and is waiting for it.
    -- Terry Pratchett, Reaper Man