[OpenAFS-devel] Re: Proposed new kafs interface: k_getpag

Thu, 16 Jul 2009 14:11:35 -0700

--Boundary_(ID_q6j52RyAZU11Wv7I3Gb4Mg)
Content-type: text/plain; charset=us-ascii; format=flowed; delsp=yes
Content-transfer-encoding: 7BIT

>
> The first thing that pops into my head is "aid in debugging". There  
> have
> been several times where I've had related processes that should be in
> the same pag and have the same credentials not be, and with some of  
> the
> weirdness we've encountered it would be nice to have the actual pag
> number so I could say "yes, both things are in the same pag[1],  
> something
> else is going on" or "no, they are not in the same pag, find out what
> caused that".
>
> And as someone else has pointed out, I'm sure someone will think of
> some reason to want the number. Why not have the interface return
> the number?

uint32_t is too small to be temporal unique in a sane way, ie you get  
to keep a list of them.

Representing pags as a number its a bad idea since that limit  
implementation options. What is the OS have native support for PAGs  
(or something kind of like it) but they are all internal and the  
consumer cant know anything about them ?

Returning a number that might be unique is ok with me, but it should  
not be the PAG number.

Love

--Boundary_(ID_q6j52RyAZU11Wv7I3Gb4Mg)
Content-type: text/html; charset=us-ascii
Content-transfer-encoding: quoted-printable

<html><head></head><body style=3D"word-wrap: break-word; =
-webkit-nbsp-mode: space; -webkit-line-break: after-white-space; =
"><br><div><blockquote type=3D"cite"><div><font class=3D"Apple-style-span"=
 color=3D"#000000"><br></font>The first thing that pops into my head is =
"aid in debugging". There have<br>been several times where I've had =
related processes that should be in<br>the same pag and have the same =
credentials not be, and with some of the<br>weirdness we've encountered =
it would be nice to have the actual pag<br>number so I could say "yes, =
both things are in the same pag[1], something<br>else is going on" or =
"no, they are not in the same pag, find out what<br>caused =
that".<br><br>And as someone else has pointed out, I'm sure someone will =
think of<br>some reason to want the number. Why not have the interface =
return<br>the number?<br></div></blockquote></div><br><div>uint32_t is =
too small to be temporal unique in a sane way, ie you get to keep a list =
of them.</div><div><br></div><div>Representing pags as a number its a =
bad idea since that limit implementation options. What is the OS have =
native support for PAGs (or something kind of like it) but they are all =
internal and the consumer cant know anything about them =
?</div><div><br></div><div>Returning a number that might be unique is ok =
with me, but it should not be the PAG =
number.</div><div><br></div><div>Love</div><div><br></div><div><br></div><=
/body></html>=

--Boundary_(ID_q6j52RyAZU11Wv7I3Gb4Mg)--