Symbolic instead of hard link as crosspost-link-function?

Symbolic instead of hard link as crosspost-link-function?

[Mikael Djurfeldt]

Is it possible to define the crosspost-link-function to use symbolic
links instead of hard links?  I can imagine that there could be
trouble due to the different semantics of symbolic links.

More precisely, will the following setting lead to trouble:

  (setq nnmail-crosspost-link-function 'make-symbolic-link)

Please respond via ordinary e-mail.  I'm not on this list.

Best regards,
/mdj

Re: Symbolic instead of hard link as crosspost-link-function?

[Michael Welsh Duggan]

Mikael Djurfeldt <mdj@nada.kth.se> writes:

> Is it possible to define the crosspost-link-function to use symbolic
> links instead of hard links?  I can imagine that there could be
> trouble due to the different semantics of symbolic links.
> 
> More precisely, will the following setting lead to trouble:
> 
>   (setq nnmail-crosspost-link-function 'make-symbolic-link)

I have been using this setting and have had no problems so far.  I
find it necessary when on a distributed file system (for obvious
reasons).

-- 
Michael Duggan               Support the anti-Spam amendment
(md5i@cs.cmu.edu)            Join at <URL:http://www.cauce.org/>

Re: Symbolic instead of hard link as crosspost-link-function?

[Samuel Tardieu]

>>>>> "Mikael" == Mikael Djurfeldt <mdj@nada.kth.se> writes:

Mikael> More precisely, will the following setting lead to trouble:
Mikael>   (setq nnmail-crosspost-link-function 'make-symbolic-link)

As directed in the documentation, you should use 'copy-file if you
cannot use hardlinks (i.e. 'add-name-to-file). The reason is that if
an article is crossposted in two mail-groups with different expiration 
times and the one with the shortest ex. time is the one containing the 
original, then you will end up with a symlink pointing to nowhere.

Mikael> Please respond via ordinary e-mail.  I'm not on this list.

Then subscribe :)

  Sam
-- 
Samuel Tardieu -- sam@ada.eu.org

Re: Symbolic instead of hard link as crosspost-link-function?

[Michael Welsh Duggan]

Samuel Tardieu <sam@ada.eu.org> writes:

> >>>>> "Mikael" == Mikael Djurfeldt <mdj@nada.kth.se> writes:
> 
> Mikael> More precisely, will the following setting lead to trouble:
> Mikael>   (setq nnmail-crosspost-link-function 'make-symbolic-link)
> 
> As directed in the documentation, you should use 'copy-file if you
> cannot use hardlinks (i.e. 'add-name-to-file). The reason is that if
> an article is crossposted in two mail-groups with different expiration 
> times and the one with the shortest ex. time is the one containing the 
> original, then you will end up with a symlink pointing to nowhere.

Ah.  I hadn't had a problem with symbolic links before, but now I
realize that that was because I never expire messages.

-- 
Michael Duggan               Support the anti-Spam amendment
(md5i@cs.cmu.edu)            Join at <URL:http://www.cauce.org/>

Re: Symbolic instead of hard link as crosspost-link-function?

[SL Baur]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

>>>>>> "MD" == Mikael Djurfeldt <mdj@nada.kth.se> writes:

MD> Is it possible to define the crosspost-link-function to use symbolic
MD> links instead of hard links?

> Important reason why not: not all filesystems support symbolic links.

Symbolic links are about 20 years old and have withstood the test of
time.  If a system hasn't implemented them by now, why can't it safely
be ignored?

Re: Symbolic instead of hard link as crosspost-link-function?

[William M. Perry]

SL Baur <steve@xemacs.org> writes:

> Stainless Steel Rat <ratinox@peorth.gweep.net> writes:
> 
> >>>>>> "MD" == Mikael Djurfeldt <mdj@nada.kth.se> writes:
> 
> MD> Is it possible to define the crosspost-link-function to use symbolic
> MD> links instead of hard links?
> 
> > Important reason why not: not all filesystems support symbolic links.
> 
> Symbolic links are about 20 years old and have withstood the test of
> time.  If a system hasn't implemented them by now, why can't it safely
> be ignored?

  Windoze.

-bp

Re: Symbolic instead of hard link as crosspost-link-function?

[Dan Schmidt]

SL Baur <steve@xemacs.org> writes:

| Stainless Steel Rat <ratinox@peorth.gweep.net> writes:
| 
| >>>>>> "MD" == Mikael Djurfeldt <mdj@nada.kth.se> writes:
| 
| MD> Is it possible to define the crosspost-link-function to use symbolic
| MD> links instead of hard links?
| 
| > Important reason why not: not all filesystems support symbolic links.
| 
| Symbolic links are about 20 years old and have withstood the test of
| time.  If a system hasn't implemented them by now, why can't it safely
| be ignored?

I have to use Windows 95 at work, which doesn't have symbolic links.
I realize that Win95 is a work of Pure Evil (I run Linux at home),
but I do manage to run Emacs and gnus on it.  I'd prefer not to be
ignored, safely or not :)

-- 
                 Dan Schmidt -> dfan@harmonixmusic.com, dfan@alum.mit.edu
Honest Bob & the                http://www2.thecia.net/users/dfan/
Factory-to-Dealer Incentives -> http://www2.thecia.net/users/dfan/hbob/
          Gamelan Galak Tika -> http://web.mit.edu/galak-tika/www/

Re: Symbolic instead of hard link as crosspost-link-function?

[Michael Lamoureux]

 "rat" == Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

>>>>>> "sb" == SL Baur <steve@xemacs.org> writes:

sb> Symbolic links are about 20 years old and have withstood the test
sb> of time.  If a system hasn't implemented them by now, why can't it
sb> safely be ignored?

rat> Windows 95, Windows/NT, MS-DOS, Macintosh, VAX\VMS, all run
rat> versions of GNU Emacs, none support symbolic or hard links.
rat> There are probably others.

If they aren't links, what are "aliases" under MacOS then?  (Not that
you can run Emacs 19 on MacOS anyway, but you brought it up.)


wondering,
Michael

Re: Symbolic instead of hard link as crosspost-link-function?

[William M. Perry]

Michael Lamoureux <lamour@engin.umich.edu> writes:

>  "rat" == Stainless Steel Rat <ratinox@peorth.gweep.net> writes:
> 
> >>>>>> "sb" == SL Baur <steve@xemacs.org> writes:
> 
> sb> Symbolic links are about 20 years old and have withstood the test
> sb> of time.  If a system hasn't implemented them by now, why can't it
> sb> safely be ignored?
> 
> rat> Windows 95, Windows/NT, MS-DOS, Macintosh, VAX\VMS, all run
> rat> versions of GNU Emacs, none support symbolic or hard links.
> rat> There are probably others.
> 
> If they aren't links, what are "aliases" under MacOS then?  (Not that
> you can run Emacs 19 on MacOS anyway, but you brought it up.)

  I hope they do a better job at it than win95/nt do.  I'd have to give my
freind jer a call to double check though.  In win95/nt, the _only_ things
that understand about aliases are the GUI components.  If you try to open
one up, you basically get an 'INI' file that tells you where the real one
is - not very bloody helpful. :)

-Bill P.

Re: Symbolic instead of hard link as crosspost-link-function?

[Michael Lamoureux]

 "rat" == Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

>>>>>> "WMP" == William M Perry <wmperry@aventail.com> writes:

>>> If they aren't links, what are "aliases" under MacOS then?

rat> Nonexistant previous to System 7.

You say that like System 7 was just released.


WMP> I hope they do a better job at it than win95/nt do.

rat> In other words, it is pretty much the same setup as Windows 95/NT.

Well, in case anyone cares, I heard it from someone (who...er...doesn't
work in the OS group at Apple ;-)  that hard links would be supported
soon.

[please permit me a moment of venting]

of course, this is the same moron who thought there were no Mac users
who actually wanted pre-emptive multi-tasking or real memory
management, and true to form, he thought hard links would just be
confusing to the user.  I wonder if the NeXT guys just giggle at the
MacOS people when they have to work together now.


for what it's worth,
Michael

Re: Symbolic instead of hard link as crosspost-link-function?

[Steve Baumgarten]

For what it's worth, aliases in MacOS work much better than symbolic
links.  For one thing, you can move the thing that's pointed to and
the alias can still be resolved.  You can also rename the the thing
that's pointed to -- and still the alias is resolved.

You can create an alias of a remote server volume, copy it to a floppy
disk, take the floppy down to the hall to another Mac (on the same
network, of course), put it in, double-click it, and mount the volume
there. 

The Mac is in need of many, many improvements, but aliases are cool
and really work much better than symbolic links. 

Steve Baumgarten
sbb@panix.com

Re: Symbolic instead of hard link as crosspost-link-function?

[Scott Blachowicz]

Steve Baumgarten <sbb@panix.com> wrote:

> For what it's worth, aliases in MacOS work much better than symbolic
> links.  For one thing, you can move the thing that's pointed to and
> the alias can still be resolved.  You can also rename the the thing
> that's pointed to -- and still the alias is resolved.

That's better?  What if you want the reference to be _symbolic_ - that is,
not to a particular thing, but to that thing's name?

Say, I have this:

    make
    make.new
    gmake -> make

then I do [the equivalent of]

    mv make make.old
    mv make.new make

Does that mean that gmake now points to make.old?

> You can create an alias of a remote server volume, copy it to a floppy
> disk, take the floppy down to the hall to another Mac (on the same
> network, of course), put it in, double-click it, and mount the volume
> there. 

Cool.

> The Mac is in need of many, many improvements, but aliases are cool
> and really work much better than symbolic links. 

They appear to work _differently_ (and probably better for most
situations).

Scott Blachowicz  Ph: 206/283-8802x240   Mathsoft (Data Analysis Products Div)
                                         1700 Westlake Ave N #500
scott@statsci.com                        Seattle, WA USA   98109
Scott.Blachowicz@seaslug.org

Re: Symbolic instead of hard link as crosspost-link-function?

[William M. Perry]

Michael Lamoureux <lamour@engin.umich.edu> writes:

[...]
> of course, this is the same moron who thought there were no Mac users who
> actually wanted pre-emptive multi-tasking or real memory management, and
> true to form, he thought hard links would just be confusing to the user.
> I wonder if the NeXT guys just giggle at the MacOS people when they have
> to work together now.

  I think hearty guffaws would be more in order than gigles.  :)

-Bill P.