[vchkpw] Opinions needed

2009-01-21 Thread Matt Brookings
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Some of you may be aware I'm working on changes and additions to the quota 
system in vpopmail.
Part of the aim of the updated system is to provide future-proofing against the 
problems we're
having now, where quota sizes and usage counts are overflowing in various 
circumstances, either
in vpopmail, or in applications utilizing vpopmail.

My question is this; would anyone ever require a quota below a megabyte, or, 
would any application
ever really need to know about specific usage counts below a megabyte?

When calculating usage, I'm thinking about making the smallest measure of unit 
a megabyte storing
the result in a 64bit unsigned integer.  The 64bit value is almost a 
requirement, but making the smallest
unit of measure a megabyte, should future-proof for quite a bit longer.
- --
/*
Matt Brookings m...@inter7.com   GnuPG Key D9414F70
Software developer Systems technician
Inter7 Internet Technologies, Inc. (815)776-9465
*/
-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFJdzVU6QgvSNlBT3ARAiFnAKCeHYx2gywPskZk5S//QBAud2VjrACfTynz
7ASJaP8uqjghUBvwbWi/SxQ=
=iOoj
-END PGP SIGNATURE-


Re: [vchkpw] Opinions needed

2009-01-21 Thread Rick Romero


I personally don't require it.  It also seems to me with disk space 
costing what it does, and compression being readily available (via 
plugins, ZFS, or what have you) it shouldn't really be an issue.


Are there any applications that display the actual quota amount that do 
NOT round to the megabyte?  It seems to me that most show percentage 
used, and round to the megabyte if they do display the actual number - 
maybe that's where the problem will lie..  Applications that convert the 
actual quota from bytes to megabytes will break.


Rick

Matt Brookings wrote:

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Some of you may be aware I'm working on changes and additions to the quota 
system in vpopmail.
Part of the aim of the updated system is to provide future-proofing against the 
problems we're
having now, where quota sizes and usage counts are overflowing in various 
circumstances, either
in vpopmail, or in applications utilizing vpopmail.

My question is this; would anyone ever require a quota below a megabyte, or, 
would any application
ever really need to know about specific usage counts below a megabyte?

When calculating usage, I'm thinking about making the smallest measure of unit 
a megabyte storing
the result in a 64bit unsigned integer.  The 64bit value is almost a 
requirement, but making the smallest
unit of measure a megabyte, should future-proof for quite a bit longer.
- --
/*
Matt Brookings m...@inter7.com   GnuPG Key D9414F70
Software developer Systems technician
Inter7 Internet Technologies, Inc. (815)776-9465
*/
-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFJdzVU6QgvSNlBT3ARAiFnAKCeHYx2gywPskZk5S//QBAud2VjrACfTynz
7ASJaP8uqjghUBvwbWi/SxQ=
=iOoj
-END PGP SIGNATURE-
  



--
Rick Romero
Need IT assistance? 
VF IT Services / VFEmail.net

www.vfit.biz / www.vfemail.net


!DSPAM:49773dc432683754956235!



Re: [vchkpw] Opinions needed

2009-01-21 Thread Wouter van der Schagt
My question is this; would anyone ever require a quota below a megabyte, 
or, would any application

ever really need to know about specific usage counts below a megabyte?


Not us, however current applications assume the format is in bytes (for 
example the maildirsize file) so calucalations will be off in some frontends 
or custom made backends when calculating the usage.


Sincerely,
- Wouter van der Schagt 



!DSPAM:49773fc232689127091031!



Re: [vchkpw] Opinions needed

2009-01-21 Thread Jukka Kurkela
i don't really see 16 exabyte mailboxes in near future. But when the 
time comes for those, there should be 128bit processors and operating 
systems, so replacing the 64bit unsigned int with a 128bit one at that 
point should not be a too great deal. Maybe future proof by defining the 
thing in one place.


just my 2 cents.

++jukka

!DSPAM:4977794732681263011309!



Re: [vchkpw] New development release

2009-01-21 Thread vchkpw

Matt Brookings wrote:

 I've tagged the current CVS HEAD as version 5.4.27, a development
 release.  It is available for download.

I've now compiled and installed on a 64 bit machine.  This is a low usage
live server, and seems to be working.  I use the CDB back end on all of my
installs.  I did not need to use any of the tricks that used to be
required to get older versions to compile on a 64 bit processor.  I
believe the 64 bit problems were fixed sometime after 5.4.18 when I
switched build system files to something created after 64 bit processors
became popular.


I have spotted one small problem...  On both of these installs the first
thing  make  did was run aclocal then run ./configure again.  I believe
this is a bug in the files created by the gnu build system.  The fix may
be to run the following:

aclocal
autoconf
autoheader
automake

before packaging the next update.  You will want to be sure you are
running with a reasonably recent version of the gnu build system.  The
test to verify it is fixed is to insure  make  doesn't run ./configure
when you are building from a fresh tarball.

This probably doesn't deserve its own release since most people will not
notice.  OTOH, someone without the build system installed will be unable
to get past the  make  step of building.

If you haven't done the svn conversion yet, you might want to fix this
before you do.  Build system file updates are messy since we have many of
the output files in cvs/svn.

Rick

!DSPAM:4977b4b832681990434868!



Re: [vchkpw] Opinions needed

2009-01-21 Thread Tom Collins

On Jan 21, 2009, at 6:46 AM, Matt Brookings wrote:
My question is this; would anyone ever require a quota below a  
megabyte, or, would any application

ever really need to know about specific usage counts below a megabyte?

When calculating usage, I'm thinking about making the smallest  
measure of unit a megabyte storing
the result in a 64bit unsigned integer.  The 64bit value is almost  
a requirement, but making the smallest
unit of measure a megabyte, should future-proof for quite a bit  
longer.


Two thoughts on quotas.

You're going to have to deal in sub-megabyte numbers, since the size  
of most messages are measured in KB.  Maybe you could track the quota  
in kbytes, rounding up/down as necessary?


All programs that deal with the quota (maildirsize file) will have to  
use 128-bit numbers (long long?) or whatever new method you come up  
with.  This includes not just vpopmail, but your IMAP server and  
potentially maildrop and qmail (if you have any .qmail files that are  
handled by qmail and not vdelivermail).  Maybe they can use a  
dynamically linked library?


-Tom


!DSPAM:4978006832684277763451!