[erlang-questions] Finding inconsistencies in record definitions and pretty printing records

[Ulf Wiger (TN/EAB)]

Joe Armstrong skrev:
> On Wed, Oct 15, 2008 at 2:38 PM, Ulf Wiger <ulf@REDACTED> wrote:
>> 2008/10/15 Thomas Lindgren <thomasl_erlang@REDACTED>:
>>> Does it handle multiple definitions of the same record name? 
>>> (#state, to take the canonical example)
> 
> Yes - it reports an error if two record declarations are not
> identical
> 
>> This is what prompted me to introduce an -export_records() 
>> directive in my exprecs contrib.
>> 
>> If I have records that I only use inside one module, I don't really
>> care if similar, but different, records are used elsewhere.
> 
> And you promise not to send a message containing a private record, 
> anywhere - and you promise that some other process will not send you
> a message with this record and you promise not to store a record in a
> data base - ....

Yes, just like I promise not to mess with data structures documented
as "opaque", or mess with contents of ets tables, just because
they're created as 'public', or send bogus messages to other
processes just because I can, or mess with other programs' mnesia
data, for that matter.

I also restrain myself from calling
rpc:multicall(nodes(), erlang,halt,[]),
even though it's perfectly possible.

I guess the question is whether your cure is worse than the
disease. ;-) I don't think it's very common that programs inadvertedly
pass around their internal data structures. If you offer a way for
people to specify some of their records as internal, they may be
more inclined to use your solution.

BR,
Ulf W