And another thing; updating broken by recent time fetching upheaval

Whilst trying to update as part of seeing if my prior ‘Rather important bug’ mail had been fixed recently, I find that it breaks after loading 51Deprecated-bf.18 and during loading Graphics-tfel.326 because "MessageNotUnderstood: Time class>>localMicrosecondClockWithOffset”. And even if was understood, it turns out that DateAndTime class>now:offset: doesn’t exist yet either.

So I’m guessing something might possibly be wonrg.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Strange OpCodes: CDC: Clear Disks and Crash

On Fri, Feb 19, 2016 at 05:04:47PM -0800, tim Rowledge wrote:
> Whilst trying to update as part of seeing if my prior ???Rather important bug???
> mail had been fixed recently, I find that it breaks after loading 51Deprecated-bf.18
> and during loading Graphics-tfel.326 because "MessageNotUnderstood:
> Time class>>localMicrosecondClockWithOffset???. And even if was understood,
> it turns out that DateAndTime class>now:offset: doesn???t exist yet either.
>

Update map problem perhaps?

> So I???m guessing something might possibly be wonrg.
>

It certainly does not seem wirght.

Dave

I confirm updating 5.1 crash.
Please fix.


On 2/19/16, 10:04 PM, "tim Rowledge" <[hidden email]> wrote:

I was able to update by, when the bug appears, press Command+Shift+S
on the code pane in the debugger to reformat the method, save it, then
proceed.

On Fri, Feb 19, 2016 at 7:04 PM, tim Rowledge <[hidden email]> wrote:

The failure seems to intermittent for me. I tried updating the same
image several times. It failed a couple of times but usually the updates
completed without error.

This probably just requires a new update map to skip over the problem
area. DateAndTime class>>nowWithOffset: was removed and added and then
moved from Kernel to 51Deprecated, so I am guessing that an update map
is need to handle this.

Dave

On Sat, Feb 20, 2016 at 09:37:25AM -0600, Chris Muller wrote:

> On 20-02-2016, at 7:57 AM, David T. Lewis <[hidden email]> wrote:
>
> The failure seems to intermittent for me. I tried updating the same
> image several times. It failed a couple of times but usually the updates
> completed without error.

How odd. I’m not at all sure how intermittent could occur with what we hoped was a deterministic update, but there we are, reality strikes again.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Strange OpCodes: HALT: No-Op

On Sat, Feb 20, 2016 at 04:25:58PM -0800, tim Rowledge wrote:
>
> > On 20-02-2016, at 7:57 AM, David T. Lewis <[hidden email]> wrote:
> >
> > The failure seems to intermittent for me. I tried updating the same
> > image several times. It failed a couple of times but usually the updates
> > completed without error.
>
> How odd. I???m not at all sure how intermittent could occur with what we hoped was a deterministic update, but there we are, reality strikes again.
>

I don't know. I've been away all day, and checking it again now I can't
get it to fail any more. Is anyone still having trouble updating their
trunk image?

Intermittent issues would not be entirely surprising in a case like this.
Anything involving Monticello is going to do a lot of DateAndTime stuff,
and there are various processes in your image that are also doing
Chronology things. If things get out of whack for a short period of
time during updating, it is certainly possible that intermittent errors
might happen.

These discussions are part of the reason that I think it will be a good
thing when we get Chronology moved into its own package. It is also the
reason that I am nervous about doing the package move, because it
introduces the possibility of totally screwing up the update stream.

Dave
 

On Sat, Feb 20, 2016 at 08:12:59PM -0500, David T. Lewis wrote:
To answer my own question: Yes, it is still broken. I tried updating trunk
beginning with a Squeak5.0-15110.image, and it fails with problems related
to #nowWithOffset:.

So there is still a problem with the update maps at this point.

Dave

On 20.02.2016, at 18:39, David T. Lewis <[hidden email]> wrote:
>
> I tried updating trunk
> beginning with a Squeak5.0-15110.image, and it fails with problems related
> to #nowWithOffset:.
>
> So there is still a problem with the update maps at this point.

I updated the config.  Could you try again? I don’t have a suitable image lying around.

- Bert -

Sadly, nope. Starting from a trunk 5.0 image as pulled out of an all-in-one it fails just the same way; DateAndTime class nowWithOffset: is sent (originating from MCPackageLoader>preambleAsCommentNamed: sending Date today) and Time class dNU: localMicrosecondClockWithOffset.

This happens during update-dtl.348.mcm


tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Useful random insult:- IQ = dx / (1 + dx), where x = age.

Ah, I got it now. The deprecated was not the actual former version. Should be fixed now?

- Bert -

> On 22-02-2016, at 3:18 PM, Bert Freudenberg <[hidden email]> wrote:
>
>
>
> Ah, I got it now. The deprecated was not the actual former version. Should be fixed now?
>
Well it worked at least once for me … thanks.


tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
The colder the X-ray table, the more of your body is required on it.

Thans for the fix. My image update to 15678


On 2/22/16, 8:18 PM, "Bert Freudenberg" <[hidden email]> wrote:

Confirming, I was able to update Squeak5.0-15113 to latest trunk.

Thanks Bert!

Dave

On Tue, Feb 23, 2016 at 07:31:44AM -0300, Edgar J. De Cleene wrote: