Re: Tests in a software release

["Randy Brukardt" <randy@rrsoftware.com>]

"Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de> wrote in message 
news:ot9cfr$1k6s$1@gioia.aioe.org...
...
> Ada could have a pragma to remove all ELSE parts of all IF-statements. 
> Would anybody argue for that? (:-))

I'd consider making an argument for a pragma to *require* an ELSE part for 
all IF statements.

The Janus/Ada style guide requires either than ELSE part or a comment 
starting -- Else before the END IF to explain why it isn't needed. We 
adopted that after a string of bugs where there was an If statement without 
an else where it appeared that no one had considered what to do if the IF 
was False. By requiring at least a comment, we make the programmer (often 
me!) think about that case.

Yes, sometimes, they end up being something like:

    -- else *** not sure what to do here!! ***

but at least that is a red flag to anyone looking at the code in a debugging 
situation.

                                Randy.