[M3devel] another Snow Leopard compiler crash

[Jay K]

That is about good enough -- your compiler is upgraded, a few things are out of date and will crash.

I'm a bit surprised you hit this.

I updated postgres95 to avoid it..but maybe that is masking a problem.

I updated upgrade.py to install a more conventional config file instead of what it used to do.
What it used to do is delegate back to cvs/m3-sys/cminstall/src/config-no-install.
Which is great if you are editing the files and want just one copy, would have saved
me a lot of time in the past, but maybe is too nichey.


Now it installs what a distribution would include -- no dependency on cvs tree.

 - Jay


----------------------------------------
> From: jay.krell at cornell.edu
> Date: Sat, 8 Jan 2011 18:50:38 -0500
> To: mika at async.caltech.edu
> CC: m3devel at elegosoft.com; jay.krell at cornell.edu
> Subject: Re: [M3devel] another Snow Leopard compiler crash
>
> Yes. When upgrade.py fails, you may or may not be in a bad state. I can see about making it safer. There are unavoidable depenedencies among cm3, cm3cg, and m3core, and maybe libm3.
>
> Starting point is a working cm3, and its cm3cg, and its config files, and m3core, and possible its libm3.
>
> I think maybe I should write a feature in upgrade.py that can start from scratch -- download old install archive or start from a specified local copy thereof. Interest? I might use it myself. I think upgrade should also have option to build all.
>
> - Jay
>
> On Jan 8, 2011, at 1:17 PM, Mika Nystrom  wrote:
>
> > Do I need to go back to the release and install that, and then upgrade.py?
> > As far as I know my compiler was working fine...
> >
> > Jay K writes:
> >> --_1def5bc1-96d7-442f-bd33-2522abadf97d_
> >> Content-Type: text/plain; charset="iso-8859-1"
> >> Content-Transfer-Encoding: quoted-printable
> >>
> >> Tony's instructions imho are too much work=2C error-prone. Upgrade.py does =
> >> rebuild cm3cg. Don't do it manually=2C as you have to be careful when you s=
> >> witch from old to new=2C and upgrade.py knows. There was a problem regardin=
> >> g m3bundle=2C fairly recently. So try head. Not a few weeks ago. The script=
> >> s specifically had a problem.
> >>
> >> Jay/phone
> >>
> >>> To: jay.krell at cornell.edu
> >>> Subject: Re: [M3devel] another Snow Leopard compiler crash=20
> >>> Date: Sat=2C 8 Jan 2011 08:10:01 -0800
> >>> From: mika at async.caltech.edu
> >>> =20
> >>> Jay K writes:
> >>>> --_8699851d-dc9a-486c-965d-eaed43f71a51_
> >>>> Content-Type: text/plain=3B charset=3D"iso-8859-1"
> >>>> Content-Transfer-Encoding: quoted-printable
> >>>>
> >>>> It no longer builds m3bundle I think. There was a problem regarding m3b=
> >> und=3D
> >>>> le for a while=3D2C because I had followed upgrade.sh too closely=3D2C w=
> >> hich ha=3D
> >>>> d this problem. And this is a sign of using old cm3cg. Are you sure you =
> >> did=3D
> >>>> n't do something weird? Uograde.py does upgrade cm3cg appropriately. Try=
> >> he=3D
> >>>> ad?
> >>>>
> >>> =20
> >>> I'm not sure I didn't do anything weird. I have a CM3 distribution that
> >>> was release=2C then I upgraded it to head as of three weeks ago using
> >>> the procedure Tony described to me a few years ago. I've been using
> >>> that with no special problems since then. I then updated to the CVS head
> >>> and ran "python upgrade.py". It does not rebuild cm3cg=2C but I can do i=
> >> t
> >>> manually and try again.
> >>> =20
> >>> Mika
> >> =
> >>
> >> --_1def5bc1-96d7-442f-bd33-2522abadf97d_
> >> Content-Type: text/html; charset="iso-8859-1"
> >> Content-Transfer-Encoding: quoted-printable
> >>
> >> 
> >> 
> >> <!--<br />> >> .hmmessage P<br />> >> {<br />> >> margin:0px=3B<br />> >> padding:0px<br />> >> }<br />> >> body.hmmessage<br />> >> {<br />> >> font-size: 10pt=3B<br />> >> font-family:Tahoma<br />> >> }<br />> >> -->
> >> 
> >> Tony's instructions imho are too much work=2C err=
> >> or-prone. Upgrade.py does rebuild cm3cg. Don't do it manually=2C as you hav=
> >> e to be careful when you switch from old to new=2C and upgrade.py knows. Th=
> >> ere was a problem regarding m3bundle=2C fairly recently. So try head. Not a=
> >> few weeks ago. The scripts specifically had a problem.

 Jay/phone
> >> r>
> To:=3B jay.krell@=3Bcornell.edu
> Subject:=3B Re:=3B=
> >> [=3BM3devel]=3B another Snow Leopard compiler crash 
> Date:=
> >> =3B Sat=2C 8 Jan 2011 08:=3B10:=3B01 -0800
> From:=3B mika@=
> >> =3Basync.caltech.edu
> 
> Jay K writes:=3B
> >=3B--_8699851d=
> >> -dc9a-486c-965d-eaed43f71a51_
> >=3BContent-Type:=3B text/plain=
> >> 9=3B charset==3B"=3Biso-8859-1"=3B
> >=3BContent-Transfer-En=
> >> coding:=3B quoted-printable
> >=3B
> >=3B It no longer build=
> >> s m3bundle I think. There was a problem regarding m3bund==3B
> >=
> >> =3Ble for a while==3B2C because I had followed upgrade.sh too closely=
> >> 1=3B2C which ha==3B
> >=3Bd this problem. And this is a sign of us=
> >> ing old cm3cg. Are you sure you did==3B
> >=3Bn'=3Bt do somethi=
> >> ng weird?=3B Uograde.py does upgrade cm3cg appropriately. Try he==3B<=
> >> br>> >=3Bad?=3B
> >=3B
> 
> I'=3Bm not sure I didn'=
> >> =3Bt do anything weird. I have a CM3 distribution that
> was release=2C=
> >> then I upgraded it to head as of three weeks ago using
> the procedure =
> >> Tony described to me a few years ago. I'=3Bve been using
> that with=
> >> no special problems since then. I then updated to the CVS head
> and r=
> >> an "=3Bpython upgrade.py"=3B. It does not rebuild cm3cg=2C but I can=
> >> do it
> manually and try again.
> 
> Mika

> >> <=
> >> /body>
> >> =
> >>
> >> --_1def5bc1-96d7-442f-bd33-2522abadf97d_--