Quoting Alan Cox <[EMAIL PROTECTED]>:

> > The absolutely worst happens when someone attempts to re-type
> > a _t over a large body of existing code, for instance inexpirienced
> > coders keep proposing to change kdev_t into a pointer.
> 
> I'd hardly call Andries inexperienced, and in fact he did precisely that
> a long time ago but the work was never integrated

I would be *very afraid* of such global changes. I saw how much damage
they can do. Several years ago, at previous job, we hired a
young-n-restless programmer. Too restless, I might add. One of his first
deeds was to redo some types and to rename some header files. He did that,
on RCS head, and went home feeling good. On the next morning all the hell
broke loose. All the other developers who worked on the head had to merge
HUNDREDS of files because their local trees became very, very different
from the head.  Those who worked on branches (and carefully chose what
files they dare to edit, for fear of merging) were shocked to learn that
their branches now include not 5 files but 500... Some of those
developers, under-caffeinated early in the morning, just checked their
local copies in, on top of changes (RCS allowed that), thus successfully
breaking *all* head builds in *many* places... It took us several weeks to
finally fix the damage, and many words were said in process...

In my opinion, the more files one proposes to change the better his
justification for that must be. The struct_t or pFoo do not break the
code, but once you start playing with them they will break something. The
Linux kernel is not in ClearCase, and developers are not looking into the
same depository. The change will render thousands of working trees
useless, and developers will need to do massive merges to catch up. It is
also important to remember that many OEMs have sources for their own
drivers - and these sources were never introduced into the kernel. Who
will be patching those, and how soon? It would be a bad idea to instantly
break all custom drivers on this planet!

Dmitri

-- 
Microsoft spel chekar vor sail, worgs grate !!
  (By [EMAIL PROTECTED], Felix von Leitner)

Attachment: msg03542/pgp00000.pgp
Description: PGP signature

Reply via email to