X-MimeOLE: Produced By Microsoft Exchange V6.5
Received: by onstor-exch02.onstor.net 
	id <01C7A3C9.B336EA45@onstor-exch02.onstor.net>; Thu, 31 May 2007 14:21:49 -0700
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Content-class: urn:content-classes:message
Subject: RE: review cw_install changes
Date: Thu, 31 May 2007 14:21:49 -0700
Message-ID: <BB375AF679D4A34E9CA8DFA650E2B04E02215809@onstor-exch02.onstor.net>
In-Reply-To: <20070531141100.30618130@ripper.onstor.net>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: review cw_install changes
Thread-Index: AcejyDCwjwgqCAJNQomMZ+FzWF5ALAAAGw1w
From: "Larry Scheer" <larry.scheer@onstor.com>
To: "Andy Sharp" <andy.sharp@onstor.com>

No, I won't make you put in an extra verification step. I used my
standalone install script to tests tar, cp and /usr/bin/install for
reliability last fall. After hundreds (maybe even a thousand or so)
passes using tar I never had a single failure. I have confidence in the
version of tar that is in our BSD.

Glad you are prepared. I think it is time this sad song comes to an end.

Later,

Larry

-----Original Message-----
From: Andy Sharp=20
Sent: Thursday, May 31, 2007 2:11 PM
To: Larry Scheer
Subject: Re: review cw_install changes

On Thu, 31 May 2007 13:11:39 -0700 "Larry Scheer"
<larry.scheer@onstor.com> wrote:

> > tar does a certain amount of verification.  is that not good enough
> for
> > you?  it's good enough for me.  there was no specification as to the
> > thoroughness of the verification.  i rest my case.
>=20
> > i could do something like this:
>=20
> > tar xf tarball 2>tarerrors
> > if [ -s tarerrors ] ; then
> >    echo gaaaaaa
> > fi
>=20
> Or how about:
>=20
> tar xf tarball 2>tarerrors
> if [ -s tarerrors ] ; then
>    echo "Gaaaack\! System fornication complete; time for a cigarette
> and a beer."
> else
>    echo "Verification completed successfully. Have a nice day"
> fi
>=20
> Just seeing if I can make you laugh....
>=20
> But, seriously...
> It will be QA or CS who will squawk about the missing verification
> phase since it is in the specification. I am just saying be prepared
> for some blow-back once it is in QA. If Sandrine is doing the QA
> work, she _will_ be looking for some form of verification success
> message.=20

I'm prepared.  Especially since I wrote the specification.  Nobody
every said how or what as to details of the verification.  So at this
point, I don't want to put in any more than I have, unless you make me.

The message is already in there, of course ~:^)

> > why don't i open a bug for all these changes and put the changes for
> > verify_install.sh towards the bugs you sent me?
>=20
> According to the notes in the defect and email I received from Raj,
> those defects were waiting for the stand-alone install.
>=20
> Opening a new defect for the silent reboots and listing the changes to
> verify_install.sh as the "fix" seems more direct. I spoke to Brian  De
> Forest and he doesn't have any problems with checking in the fix with
> or without a defect. Having a defect for this specific issue may help
> getting it in 2.2.3.
>=20
> Later
>=20
> Larry
>=20
> -----Original Message-----
> From: Andy Sharp=20
> Sent: Wednesday, May 30, 2007 5:51 PM
> To: Larry Scheer
> Subject: Re: review cw_install changes
>=20
> On Wed, 30 May 2007 17:06:59 -0700 "Larry Scheer"
> <larry.scheer@onstor.com> wrote:
>=20
> > Makefile comments:
> > Write cw_install.sh to $(PATH_TO_RELEASE)/etc .vs Tools. Don't
> > pollute source trees with derived objects. Then you don't need the
> > move.
>=20
> fixed
>=20
> > Line 562: Looking at how N_VERS is used in cw_install, you should
> > use the make macro $(VER) which is the contents of
> > nfx-tree/Tools/version. It doesn't have the 'b' character appended
> > to it. So it will just be 3.0.0.0 or 2.2.3.0, etc. The version file
> > in the release directory has a character 'b' appended to it to
> > differentiate between bobcat and cheetah. (Cheetah's version file
> > does not have the 'b'. I have no idea what it will be for cougar
> > and bobcat linux.
>=20
> fixed
>=20
> > Cw_install.in comments:
> > Line 106: hw_initials for Bobcat Linux will most likely be 'BL'
>=20
> fixed.  buttload.  should be BC  gaaaah
>=20
> > Line: 582: Missing preposition before "most cases." Perhaps is
> > should be "in most cases." Or "for most cases."  (Whatever is your
> > preferences (or most gooder English.)
>=20
> ficked
>=20
> > Line 596-597: Maybe it should say: "After a successful upgrade, your
> > system will boot the software and you will have to log in again."
> > Or it could be "When efficacious upgradation consummated system
> > reboot will occur, necessitating login procedures upon completion
> > of init level two."
>=20
> uh
>=20
> > Line 640: htype=3D`hw_initials` I think you meant =
hwtype=3D`hw_initials`
> > because htype is unused.
>=20
> fixed
>=20
> > Lines 644-653: Unless you change the Makefile as mentioned above you
> > can't use $N_VERS unmodified for anything but Cheetah builds. For
> > bobcat (BSD) N_VERS would be set to 3.0.0.0b which breaks all of
> > your default URLs.
>=20
> I said fixed already
>=20
> > The functional specification for the CW installation procedure has a
> > requirement that the script "verifies the integrity of the copied
> > files" (Pg. 9 of 14) and "verify the install before performing the
> > reboot of the installed release." (Pg. 6 of 14) I did not see where
> > the script did this.
>=20
> tar does a certain amount of verification.  is that not good enough
> for you?  it's good enough for me.  there was no specification as to
> the thoroughness of the verification.  i rest my case.
>=20
> i could do something like this:
>=20
> tar xf tarball 2>tarerrors
> if [ -s tarerrors ] ; then
>     echo gaaaaaa
> fi
>=20
> > Changes to flash_install.sh can go with this change list or a
> > separate one. It does make sense to check the file in with this
> > change list.
>=20
> yeah, they're just comments
>=20
> > Verify_install.sh can go with this change, but I think a bug should
> > be opened and filed for this fix and add this fix the Delorean
> > release. It definitely should be back-ported to 2.2.3.0. Assigning
> > a defect to it would expedite the process. (I would hope anyway.)
> > Brian De Forest should be made aware of this fix.
>=20
> why don't i open a bug for all these changes and put the changes for
> verify_install.sh towards the bugs you sent me?
>=20
