[WBEL-devel] An "Issue" with WBEL (pun intended)

[John A. Tamplin]

On 13 Dec 2003, Simon J Mudd wrote:

> This is something which I think is an important issue to address.
> How similar to RH does WBL want to be?
> 
> I understand the need to distinguish the product from RHEL3 but I
> think it is a mistake if applications have to do special things to
> recognise the OS.

To me, this is exactly analogous to the browser user agent strings, where
everything claims to be some version of Mozilla.  Clearly a better way to
determine what to do for the OS you are installing on is to detect
particular features rather than the name.  Ie, if you need to do something
different based on the version of glibc, look at the version of glibc.  
Likewise for other aspects of the distribution, such as the filesystem
structure.

> It would be better IMO that the redhat-release package mimicked the RH
> version, and perhaps whatever other tests are generally will work
> equally well on WBL.  That way applications designed for RHEL should
> recognise WBL as RHEL and should generally work without
> problems... AND vice versa. This provides a larger audience to the
> software for both OSes.

I think this would be a very gray area legally -- you are essentially 
claiming that the OS *is* RHEL.

> However it would be handy to have an officially recognised way of
> checking if the OS is WBL, and determining its release.  That way all
> applications can consistently perform the same check and determine
> which version of WBL is being used.  Whether this is to check the
> output of rpm -q whitebox-release, or look at /etc/whitebox-release
> does not really matter it is the need for a consistent test which can
> be used if necessary.
> 
> Most other RPM based Linuxes do not appear to be consistent with this
> and this makes it complicated determining on which OS an application
> is running.

But does it really matter which OS?  What an application cares about is
what needs to be done to make it function.  IMHO, an application that does
so based on the specific text name of the OS is needlessly restrictive and
either requires an enormous database of all the names (which would be a
massive maintenance nightmare) it could find or will run on a tiny subset
of the OSes that it otherwise could run on.

-- 
John A. Tamplin					jat@jaet.org
770/436-5387 HOME				4116 Manson Ave
                		 		Smyrna, GA  30082-3723