[rrd-developers] it exposes too much ...

Tobias Oetiker tobi at oetiker.ch
Tue Jun 10 14:25:53 CEST 2008

Hi Florian,

Today Florian Forster wrote:

> Hi again,
> On Tue, Jun 10, 2008 at 01:20:44PM +0200, Tobias Oetiker wrote:
> > ..., and export-documenting the bits that some people may be using.
> so, if I understand you correctly, there will be functions in 1.3 that
> are exported but not documented and that will be removed in 1.4. Right?
> How is the situation of the people who need these functions improved by
> not documenting them? If the functions are removed in 1.4 they will need
> to adapt - whether the functions are declared ``officially'' or not.
they will not be removed just like that, they will be replaced by
'good functions' with the same intention as the current non
supported / non documented functions. In other words, not exporting
the functions now, would cause the users to have no way but
duplicating rrdtool code somehow or compiling an alternete version
... in 1.4 they will have the option to switch to the supported
functions ...

> How is librrd's situation improved by not declaring those functions
> externally? Are you limited in your plans for 1.4 in any way?
> I guess you're concerned about publishing functions that will be removed
> in the foreseeable future. But not publishing doesn't change the fact
> that some people (might) use them. If you rather export the functions
> than not, why not declare them using __attribute__((deprecated))? This
> way you can show people explicitly that they shouldn't use those
> functions (instead of showing them implicitly by hiding the functions).
> In the best case people don't care about exported but undocumented
> functions. In all other cases exported but undocumented functions are
> worse than exported and documented functions.

that  __attribute__((deprecated)) thing looks cool. How does it
work ? can you send a patch ?

There is another issue with documenting the 'exported /
non-supported' functions, they depend on rrd-format.h which means
that rrd-format.h would have to become an exported header which is
realy not what I intend ...

> Again, just my $.02 - I just didn't have the impression I got my point
> across..

I think I understand your intention. You want things (however they
are) in the open.


> Regards,
> -octo

Tobi Oetiker, OETIKER+PARTNER AG, Aarweg 15 CH-4600 Olten
http://it.oetiker.ch tobi at oetiker.ch ++41 62 213 9902

More information about the rrd-developers mailing list