Re: #774: [PATCH] avoid early exit when acting on some non-versioned files

Karl Fogel wrote:

>Eric Gillespie <epg@pretzelnet.org> writes:
>
>
>>>Our practice has been, if some caller tests for a specific error, then
>>>that particular error should be documented in the callee, and all the
>>>way down as necessary.
>>>
>>>
>>That's a pretty haphazard way to write documentation.
>>
>>
>
>Sure, it would certainly be better for every function to document
>every error code it could return. If we could do it over again,
>that's probably what we'd do. This solution is a compromise; I never
>said it wasn't.
>
It's also practically impossible to maintain such documentation. Any new
error code produced by a function would have to be documented in all its
callers, recursively. I've lived on projects that expected things to be
documented like that, and I haven't seen a single case where this rule
was consistenly adhered to. Bits rot much, much faster if you impose an
impossible rule.

Now, if we had a script that automated such docs, things would be different.

-- 
Brane Čibej   <brane_at_xbc.nu>   http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org