Hardlink Pitfalls (was: Patches for REALLY TINY 386 kernels)

Hardlink Pitfalls (was: Patches for REALLY TINY 386 kernels)

[Al Boldi]

Satyam Sharma wrote:
> Or just "cp -al" to create multiple trees at (almost) no disk cost
> that won't interfere with each other in any way, and makes the
> development process / generating patchsets trifle easier as well ...

That would be correct if hardlinks would actually do a CoW on modify, instead 
of misleading the user into thinking he is modifying an independent file.

Moral of the story:  try to avoid hardlinks as much as possible!


Thanks!

--
Al

Re: Hardlink Pitfalls (was: Patches for REALLY TINY 386 kernels)

[Satyam Sharma]

On 7/16/07, Al Boldi <a1426z@gawab.com> wrote:
> Satyam Sharma wrote:
> > Or just "cp -al" to create multiple trees at (almost) no disk cost
> > that won't interfere with each other in any way, and makes the
> > development process / generating patchsets trifle easier as well ...
>
> That would be correct if hardlinks would actually do a CoW on modify, instead
> of misleading the user into thinking he is modifying an independent file.

"patch" already handles this correctly ... and backupcopy=auto,breakhardlink
in vim (and something similar in other editors I presume)

Satyam

Re: Hardlink Pitfalls (was: Patches for REALLY TINY 386 kernels)

[Jörn Engel]

On Mon, 16 July 2007 22:14:41 +0530, Satyam Sharma wrote:
> On 7/16/07, Al Boldi <a1426z@gawab.com> wrote:
> >Satyam Sharma wrote:
> >> Or just "cp -al" to create multiple trees at (almost) no disk cost
> >> that won't interfere with each other in any way, and makes the
> >> development process / generating patchsets trifle easier as well ...
> >
> >That would be correct if hardlinks would actually do a CoW on modify, 
> >instead
> >of misleading the user into thinking he is modifying an independent file.
> 
> "patch" already handles this correctly ... and backupcopy=auto,breakhardlink
> in vim (and something similar in other editors I presume)

Yet even a single program failing to break the link will leave you with
a mess to sort out.  The only place that can ensure to always break the
link is the kernel.  Which is why I wrote the cowlink patches some years
back.

The still need a lot of love to be merge-ready.  But I do use them on a
daily basis.

Jörn

-- 
Joern's library part 2:
http://www.art.net/~hopkins/Don/unix-haters/tirix/embarrassing-memo.html
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Hardlink Pitfalls (was: Patches for REALLY TINY 386 kernels)

[Al Boldi]

Jörn Engel wrote:
> On Mon, 16 July 2007 22:14:41 +0530, Satyam Sharma wrote:
> > On 7/16/07, Al Boldi <a1426z@gawab.com> wrote:
> > >Satyam Sharma wrote:
> > >> Or just "cp -al" to create multiple trees at (almost) no disk cost
> > >> that won't interfere with each other in any way, and makes the
> > >> development process / generating patchsets trifle easier as well ...
> > >
> > >That would be correct if hardlinks would actually do a CoW on modify,
> > >instead
> > >of misleading the user into thinking he is modifying an independent
> > > file.
> >
> > "patch" already handles this correctly ... and
> > backupcopy=auto,breakhardlink in vim (and something similar in other
> > editors I presume)
>
> Yet even a single program failing to break the link will leave you with
> a mess to sort out.

No kidding!

> The only place that can ensure to always break the
> link is the kernel.  Which is why I wrote the cowlink patches some years
> back.

Can you post a patch against 2.6.22?

> The still need a lot of love to be merge-ready.  But I do use them on a
> daily basis.

Well, if the patch is lean and optional and runtime configurable, then they 
should definitely be considered for inclusion.


Thanks!

--
Al

-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Hardlink Pitfalls (was: Patches for REALLY TINY 386 kernels)

[Jörn Engel]

On Mon, 16 July 2007 20:23:04 +0300, Al Boldi wrote:
> Jörn Engel wrote:
> 
> > The only place that can ensure to always break the
> > link is the kernel.  Which is why I wrote the cowlink patches some years
> > back.
> 
> Can you post a patch against 2.6.22?

I can and probably will.

> > The still need a lot of love to be merge-ready.  But I do use them on a
> > daily basis.
> 
> Well, if the patch is lean and optional and runtime configurable, then they 
> should definitely be considered for inclusion.

Lean is not the word I would choose.  It works.  For me.  That is about
as much praise as it deserves.

Jörn

-- 
Beware of bugs in the above code; I have only proved it correct, but
not tried it.
-- Donald Knuth
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Hardlink Pitfalls (was: Patches for REALLY TINY 386 kernels)

[Geert Uytterhoeven]

On Mon, 16 Jul 2007, Al Boldi wrote:
> Satyam Sharma wrote:
> > Or just "cp -al" to create multiple trees at (almost) no disk cost
> > that won't interfere with each other in any way, and makes the
> > development process / generating patchsets trifle easier as well ...
> 
> That would be correct if hardlinks would actually do a CoW on modify, instead 
> of misleading the user into thinking he is modifying an independent file.
> 
> Moral of the story:  try to avoid hardlinks as much as possible!

Or use a different user account.

I used to have `all' Linux kernel source trees hardlinked where possible, as
special user `src'. So I could create a `cp -rl' copy under my own account and
get an error when trying to modify a file.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds