Darren.Reed at Sun.COM wrote:
> Bill Sommerfeld wrote:
>
>> ...
>> We are only requesting permission to allocate the names, integrate the
>> platmods, and establish the directories for them in /platform and
>> /usr/platform at this time. We will use symbolic links to existing
>> platforms or the generic sun4u directories where appropriate.
>>
>> ...
>> We would like to use the code names above in the source tree, so that
>> we would be adding directories such as "usr/src/uts/sun4u/douglas" and
>> "usr/src/uts/sun4u/phoenix" to the tree.
>>
>
> Has any consideration been given to using the prefix (TAD)
> as either a directory or part of the name of the directory in
> usr/src? While we don't have a large number of competitors
> here, it would seem reasonable to provide/enforce the means
> for multiple vendors to use the same "code name" without
> stepping on each other's toes for directories.
>
> Whilst I'm aware that there are presently only two vendors
> active in this space (and not likely to be many more), it
> would seem we can benefit.
I don't mind having TAD, as a directory prefix, but I do not want to
create a separate hierarchy, for the simple reason that having the
directories live at different levels will introduce gratuitous change in
Makefiles that are otherwise nearly identical. (E.g. sparcle has
Makefiles that are copied from grover, and are almost identical.)
At this point, I consider the likelihood of directory name collisions to
be really small though, and the extra TAD, prefix on some directories
will have some ugly effects:
1) it will group the Tadpole platforms together -- AT THE TOP OF THE
LISTING -- just after the Makefiles.
2) because of the above, it may imply that there is something "special"
about either those names, or the platforms (and indeed there is, in so
much as the fact that they are not Sun produced.)
3) it will just "look" ugly in the directory listings.
Now, if we were to rename "all" the platform directories, such that
SUNW,xxx is there next to TAD,xxx, that might work out ok. But then I
would propose to abandon the code names in the directories, and instead
use the actual uname -i's as the directory names. E.g. TAD,SPARCE, or
SUNW,Ultra-5_10. :-) That would certainly reduce on layer of
obfuscation in the source tree. :-)
-- Garrett
>
> Darren
>
