Re: strtonum(3) from OpenBSD?

[Marc Balmer <marc%msys.ch@localhost>]

Thor,

> On Tue, Jun 23, 2009 at 05:21:49PM +0200, Marc Balmer wrote:
> > Am 23.06.2009 um 17:15 schrieb Thor Lancelot Simon:
> >
> > > On Tue, Jun 23, 2009 at 05:06:29PM +0200, Marc Balmer wrote:
> > > > Yes, already more than one program apparently use it.  And at least
> > > > one
> > > > more once we have it ;)
> > >
> > > Oh, my god, three programs use this bogus interface.  That by itself
> > > must be a reason to check it in!
> >
> > care to elaborate a bit (on a technical level, if possible) why you  
> > find
> > this interface bogus?

thank you for elaborating a bit on why you called strtonum(3) bogus.
I have added my comments inline.

> Returning an error *string* is so profoundly bogus I have trouble
> imagining that anyone with more than a few weeks' experience  
> programming
> in C doesn't see why.

with even more experience in C you get to learn to actually look
at implementations.  In the strtonum(3) case, you will spot that
it actually sets errno _and_ provides a string value which can be
used for most (simple and english) error message.

> 1) You can't switch on an error string, so you can't really ever
>   programmatically do anything smart in response to failure.

Why on earth to want to switch on the error string when strtonum(3)
comfortably sets errno?

> 2) If the error string is made a defined part of the interface, so  
> that
>   programs can "switch" on it using strcmp (UGH!), then the function
>   becomes inherently non-localizable.

Why on earth to want to switch on the error string using strcmp when
strtonum(3) comfortably sets errno?

> 3) We already have defined errno values for all the error cases this
>   function can reasonably be expected to encounter if properly  
> written.

The function is properly written, I think.  These guys who wrote it
were not amateurs either.  But you failed to look at the actual
implementation...

> "OpenBSD did it" is no more reason to dirty up our libc with this kind
> of ill-considered interface than "SCO Xenix did it".  NetBSD libc  
> should
> not be the union of every mistake anyone ever made in libc in any  
> version
> of Unix.

I *NEVER* mentioned the OpenBSD origin as a *REASON* to have it.
I just asked for a technical reasoning why not have a function that is
used in several places in a library, and ideally in the same library as
other OS'es put it.  I think that is a reasonable question.

<mirror>
Oh, you use strong words, given that you did not even look at the
implementation.

You call it "dirt", "a mistake", "ill considered", do a comparison with
Xenix, wihtout even looking at the code in question...  How can I
classify your arguments as anything other than an uninformed rant?
</mirror>

Si tacuisses.... ;)

- Marc Balmer

> --
> Thor Lancelot Simon	                                    
> tls%rek.tjls.com@localhost
>    "Even experienced UNIX users occasionally enter rm *.* at the UNIX
>     prompt only to realize too late that they have removed the wrong
>     segment of the directory structure." - Microsoft WSS whitepaper