> > This change is dangerous: genName is also used in tests different from the 
> > ones 
> > you added, in particular in ones used to verify encoding and decoding; so 
> > using 
> > only those specific names makes those exisiting tests less useful. What is 
> > the 
> > purpose of that change?
> 
> 
> I've been running the test suite multiple times as I was changing the tests 
> and I didn't see any failures in the rest of the test cases. Is this really 
> that dangerous?

I meant it is dangerous in the sense of testing less(!). So the risk that I'm
seeing is that you might miss bugs that later changes introduce, e.g., to the
serialisation code that otherwise would have been caught by this test.

Basically, your change is like removing test cases; all the resulting test suits
will, of course, pass, but it is still dangerous, as you the completeness of the
tests is reduced.

> What kind of troubles for decoding/encoding can this change introduce?

You will no longer test if you correctly encode/decode host names that are more
complicated than "name1" to "name40". While the other considered characters in 
host
names (minus sign, underscore) require no special quoting in JSON, it is still 
a good
idea to test whether they get encoded correctly; same for FQDN of different 
length
or with duplicate parts (like foo.foo.example.com), etc.

Regards,
Klaus

-- 
Klaus Aehlig
Google Germany GmbH, Erika-Mann-Str. 33, 80636 Muenchen
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg
Geschaeftsfuehrer: Matthew Scott Sucherman, Paul Terence Manicle

Reply via email to