Smalltalk › Squeak › Squeak - Dev

The Trunk: Chronology-Core-dtl.2.mcz

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

12 messages Options

commits-2

The Trunk: Chronology-Core-dtl.2.mcz

David T. Lewis uploaded a new version of Chronology-Core to project The Trunk:
http://source.squeak.org/trunk/Chronology-Core-dtl.2.mcz

==================== Summary ====================

Name: Chronology-Core-dtl.2
Author: dtl
Time: 28 February 2016, 8:24:12.86775 pm
UUID: 0dc832c9-bf4c-4559-a705-c19f44392089
Ancestors: Chronology-Core-dtl.1, Kernel-tfel.1001

Re-save Chronology-Core-dtl.1 with Kernel-tfel.1001 added to ancestry to preserve version history from Kernel package.

=============== Diff against Chronology-Core-dtl.1 ===============

David T. Lewis

Re: The Trunk: Chronology-Core-dtl.2.mcz

This was my attempt to attach the prior version history for Chronology to the
new Chronology-Core and Chronology-Tests packages. It appears to break the update
stream, so I will delete Chronology-Core-dtl.2 and Chronology-Tests-dtl.2 updates
from the source.squeak.org repository.

Hopefully no one has tried to update their trunk images during this unfortunate
experiment.

Sorry,
Dave

On Mon, Feb 29, 2016 at 02:05:19AM +0000, [hidden email] wrote:

> David T. Lewis uploaded a new version of Chronology-Core to project The Trunk:
> http://source.squeak.org/trunk/Chronology-Core-dtl.2.mcz
>
> ==================== Summary ====================
>
> Name: Chronology-Core-dtl.2
> Author: dtl
> Time: 28 February 2016, 8:24:12.86775 pm
> UUID: 0dc832c9-bf4c-4559-a705-c19f44392089
> Ancestors: Chronology-Core-dtl.1, Kernel-tfel.1001
>
> Re-save Chronology-Core-dtl.1 with Kernel-tfel.1001 added to ancestry to preserve version history from Kernel package.
>
> =============== Diff against Chronology-Core-dtl.1 ===============
>

David T. Lewis

Preserving package history while refactoring a package (was: The Trunk: Chronology-Core-dtl.2.mcz)

If I want to move a sub-package out of a larger package (for example move
Kernel-Chronology out of package Kernel and into its own package called
Chronology-Core), and if I want to preserve the version history for that
sub-package, is there a way to do this?

I do not see any tool support for it. I tried manually updating the ancestry
for the new package (Chronology-Core) and adding the ancestry from a suitable
Kernel ancestor. This seems to produce a resonable version history in my
local image, and all looked good when I saved the package to local package-cache.

After copying the locally saved MCZ's with added version history up to
source.squeak.org, the result was a broken update stream. So ....

My manual hack is clearly not the right way to attack this problem. Is
there another approach that might work?

I really do not want to lose version history on the entire Chronology
package, so I am hoping that there is some simple thing that I may have
overlooked.

Thanks,
Dave

On Sun, Feb 28, 2016 at 09:17:42PM -0500, David T. Lewis wrote:

> This was my attempt to attach the prior version history for Chronology to the
> new Chronology-Core and Chronology-Tests packages. It appears to break the update
> stream, so I will delete Chronology-Core-dtl.2 and Chronology-Tests-dtl.2 updates
> from the source.squeak.org repository.
>
> Hopefully no one has tried to update their trunk images during this unfortunate
> experiment.
>
> Sorry,
> Dave
>
>
> On Mon, Feb 29, 2016 at 02:05:19AM +0000, [hidden email] wrote:
> > David T. Lewis uploaded a new version of Chronology-Core to project The Trunk:
> > http://source.squeak.org/trunk/Chronology-Core-dtl.2.mcz
> >
> > ==================== Summary ====================
> >
> > Name: Chronology-Core-dtl.2
> > Author: dtl
> > Time: 28 February 2016, 8:24:12.86775 pm
> > UUID: 0dc832c9-bf4c-4559-a705-c19f44392089
> > Ancestors: Chronology-Core-dtl.1, Kernel-tfel.1001
> >
> > Re-save Chronology-Core-dtl.1 with Kernel-tfel.1001 added to ancestry to preserve version history from Kernel package.
> >
> > =============== Diff against Chronology-Core-dtl.1 ===============
> >

Chris Muller-3

Re: The Trunk: Chronology-Core-dtl.2.mcz

In reply to this post by David T. Lewis

Hi Dave, why did attaching teh history break the update stream?

Incidentlally, the update stream was _already_ broken before, and is
still is, due to DateAndTime>>#< using an underscore assignment...

On Sun, Feb 28, 2016 at 8:17 PM, David T. Lewis <[hidden email]> wrote:

> This was my attempt to attach the prior version history for Chronology to the
> new Chronology-Core and Chronology-Tests packages. It appears to break the update
> stream, so I will delete Chronology-Core-dtl.2 and Chronology-Tests-dtl.2 updates
> from the source.squeak.org repository.
>
> Hopefully no one has tried to update their trunk images during this unfortunate
> experiment.
>
> Sorry,
> Dave
>
>
> On Mon, Feb 29, 2016 at 02:05:19AM +0000, [hidden email] wrote:
>> David T. Lewis uploaded a new version of Chronology-Core to project The Trunk:
>> http://source.squeak.org/trunk/Chronology-Core-dtl.2.mcz
>>
>> ==================== Summary ====================
>>
>> Name: Chronology-Core-dtl.2
>> Author: dtl
>> Time: 28 February 2016, 8:24:12.86775 pm
>> UUID: 0dc832c9-bf4c-4559-a705-c19f44392089
>> Ancestors: Chronology-Core-dtl.1, Kernel-tfel.1001
>>
>> Re-save Chronology-Core-dtl.1 with Kernel-tfel.1001 added to ancestry to preserve version history from Kernel package.
>>
>> =============== Diff against Chronology-Core-dtl.1 ===============
>>
>

Chris Muller-3

Re: Preserving package history while refactoring a package (was: The Trunk: Chronology-Core-dtl.2.mcz)

In reply to this post by David T. Lewis

Seems like Adopt should work...

On Sun, Feb 28, 2016 at 8:32 PM, David T. Lewis <[hidden email]> wrote:

> If I want to move a sub-package out of a larger package (for example move
> Kernel-Chronology out of package Kernel and into its own package called
> Chronology-Core), and if I want to preserve the version history for that
> sub-package, is there a way to do this?
>
> I do not see any tool support for it. I tried manually updating the ancestry
> for the new package (Chronology-Core) and adding the ancestry from a suitable
> Kernel ancestor. This seems to produce a resonable version history in my
> local image, and all looked good when I saved the package to local package-cache.
>
> After copying the locally saved MCZ's with added version history up to
> source.squeak.org, the result was a broken update stream. So ....
>
> My manual hack is clearly not the right way to attack this problem. Is
> there another approach that might work?
>
> I really do not want to lose version history on the entire Chronology
> package, so I am hoping that there is some simple thing that I may have
> overlooked.
>
> Thanks,
> Dave
>
>
>
> On Sun, Feb 28, 2016 at 09:17:42PM -0500, David T. Lewis wrote:
>> This was my attempt to attach the prior version history for Chronology to the
>> new Chronology-Core and Chronology-Tests packages. It appears to break the update
>> stream, so I will delete Chronology-Core-dtl.2 and Chronology-Tests-dtl.2 updates
>> from the source.squeak.org repository.
>>
>> Hopefully no one has tried to update their trunk images during this unfortunate
>> experiment.
>>
>> Sorry,
>> Dave
>>
>>
>> On Mon, Feb 29, 2016 at 02:05:19AM +0000, [hidden email] wrote:
>> > David T. Lewis uploaded a new version of Chronology-Core to project The Trunk:
>> > http://source.squeak.org/trunk/Chronology-Core-dtl.2.mcz
>> >
>> > ==================== Summary ====================
>> >
>> > Name: Chronology-Core-dtl.2
>> > Author: dtl
>> > Time: 28 February 2016, 8:24:12.86775 pm
>> > UUID: 0dc832c9-bf4c-4559-a705-c19f44392089
>> > Ancestors: Chronology-Core-dtl.1, Kernel-tfel.1001
>> >
>> > Re-save Chronology-Core-dtl.1 with Kernel-tfel.1001 added to ancestry to preserve version history from Kernel package.
>> >
>> > =============== Diff against Chronology-Core-dtl.1 ===============
>> >
>

David T. Lewis

Re: Preserving package history while refactoring a package (was: The Trunk: Chronology-Core-dtl.2.mcz)

I know how to adopt a different ancestor within a package, but I don't see
any obvious way to adopt an ancestor from a different package.

Don't laugh, but what I actually did was use an inspector to find the
ancestry for the version that I wanted to adopt from Kernel, saved its
ancestors array with "Smalltalk at: #FOO put ancestors", then in an
inspector for the version in Chronology-Core, I did "ancestors :=
ancestors, FOO". After saving this to my package cache, I had an MCZ with
plausible looking ancestry. I did the same for Chronology-Tests.

I then copied the new MCZs to trunk, and tried updating an image. The
process seemed to be stuck running "forever", though I am not sure why. I
tried it a couple of times with the same result.

At that point I quickly decided to delete the new MCZs before I really
screwed something up badly in trunk. I figured that it was time to ask for
help before I really made a mess of things.

Dave

> Seems like Adopt should work...
>
> On Sun, Feb 28, 2016 at 8:32 PM, David T. Lewis <[hidden email]>
> wrote:
>> If I want to move a sub-package out of a larger package (for example
>> move
>> Kernel-Chronology out of package Kernel and into its own package called
>> Chronology-Core), and if I want to preserve the version history for that
>> sub-package, is there a way to do this?
>>
>> I do not see any tool support for it. I tried manually updating the
>> ancestry
>> for the new package (Chronology-Core) and adding the ancestry from a
>> suitable
>> Kernel ancestor. This seems to produce a resonable version history in my
>> local image, and all looked good when I saved the package to local
>> package-cache.
>>
>> After copying the locally saved MCZ's with added version history up to
>> source.squeak.org, the result was a broken update stream. So ....
>>
>> My manual hack is clearly not the right way to attack this problem. Is
>> there another approach that might work?
>>
>> I really do not want to lose version history on the entire Chronology
>> package, so I am hoping that there is some simple thing that I may have
>> overlooked.
>>
>> Thanks,
>> Dave
>>
>>
>>
>> On Sun, Feb 28, 2016 at 09:17:42PM -0500, David T. Lewis wrote:
>>> This was my attempt to attach the prior version history for Chronology
>>> to the
>>> new Chronology-Core and Chronology-Tests packages. It appears to break
>>> the update
>>> stream, so I will delete Chronology-Core-dtl.2 and
>>> Chronology-Tests-dtl.2 updates
>>> from the source.squeak.org repository.
>>>
>>> Hopefully no one has tried to update their trunk images during this
>>> unfortunate
>>> experiment.
>>>
>>> Sorry,
>>> Dave
>>>
>>>
>>> On Mon, Feb 29, 2016 at 02:05:19AM +0000, [hidden email]
>>> wrote:
>>> > David T. Lewis uploaded a new version of Chronology-Core to project
>>> The Trunk:
>>> > http://source.squeak.org/trunk/Chronology-Core-dtl.2.mcz
>>> >
>>> > ==================== Summary ====================
>>> >
>>> > Name: Chronology-Core-dtl.2
>>> > Author: dtl
>>> > Time: 28 February 2016, 8:24:12.86775 pm
>>> > UUID: 0dc832c9-bf4c-4559-a705-c19f44392089
>>> > Ancestors: Chronology-Core-dtl.1, Kernel-tfel.1001
>>> >
>>> > Re-save Chronology-Core-dtl.1 with Kernel-tfel.1001 added to ancestry
>>> to preserve version history from Kernel package.
>>> >
>>> > =============== Diff against Chronology-Core-dtl.1 ===============
>>> >
>>
>

Bert Freudenberg

Re: The Trunk: Chronology-Core-dtl.2.mcz

In reply to this post by Chris Muller-3

On 29.02.2016, at 16:55, Chris Muller <[hidden email]> wrote:
>
> Incidentally, the update stream was _already_ broken before, and is
> still is, due to DateAndTime>>#< using an underscore assignment…

What? Didn’t I fix that in Kernel-bf.998?

Btw this came from me enabling the syntaxHighlightingAsYouTypeLeftArrowAssignment preference assuming that it would show assignment arrows but store as colon-equals. Maybe we should get rid of that, or fix it to do just that.

- Bert -

smime.p7s (5K) Download Attachment

Chris Muller-3

Re: The Trunk: Chronology-Core-dtl.2.mcz

On Tue, Mar 1, 2016 at 8:17 AM, Bert Freudenberg <[hidden email]> wrote:
> On 29.02.2016, at 16:55, Chris Muller <[hidden email]> wrote:
>>
>> Incidentally, the update stream was _already_ broken before, and is
>> still is, due to DateAndTime>>#< using an underscore assignment…
>
> What? Didn’t I fix that in Kernel-bf.998?

I'm referring to the update stream being broken. A stock
Squeak5.0-15113 cannot currently be updated to latest trunk because of
that underscore. I wanted to make sure Dave wasn't confusing that
with his change.

> Btw this came from me enabling the preference assuming that it would show assignment arrows but store as colon-equals. Maybe we should get rid of that, or fix it to do just that.

+1

Chris Muller-3

Re: The Trunk: Chronology-Core-dtl.2.mcz

> I'm referring to the update stream being broken. A stock
> Squeak5.0-15113 cannot currently be updated to latest trunk because of
> that underscore. I wanted to make sure Dave wasn't confusing that
> with his change.

Correction: Okay, I didn't realize we allowUnderscoreAssignments by
default. Really, still?

When I make my trunk image, I have been loading my own prefs to set my
updateUrl to trunk for myself, but it also includes setting this off,
and so that's why it was blowing up for me.

Bert Freudenberg

Re: The Trunk: Chronology-Core-dtl.2.mcz

> On 01.03.2016, at 21:15, Chris Muller <[hidden email]> wrote:
>
>> I'm referring to the update stream being broken. A stock
>> Squeak5.0-15113 cannot currently be updated to latest trunk because of
>> that underscore. I wanted to make sure Dave wasn't confusing that
>> with his change.
>
> Correction: Okay, I didn't realize we allowUnderscoreAssignments by
> default. Really, still?
>
> When I make my trunk image, I have been loading my own prefs to set my
> updateUrl to trunk for myself, but it also includes setting this off,
> and so that's why it was blowing up for me.

Yes we should change that pref.

And it looks like the differ replaces underscore assignments by ansi assignments. Maybe that’s why I missed one line when publishing my fix ...

Not entirely sure how to fix it since various subsequent versions of Kernel have been committed, and then the code has been moved to Chronology. Well, we at least should change the offending line in Chronology now. It’s in DateAndTime>><.

Btw I’ve been thinking about how to get a “proper” ancestry of Chronology and I guess the right thing would be to write a script that goes through all Kernel package versions, carves out a Kernel-Chronology package from it, and commits that. It could filter versions where no definition in Kernel-Chronology changed. The latest of those packages could then be adopted as ancestor of Chronology-Core.

I don’t think we ever have done that kind of history surgery in Monticello, but it is something I’ve heard people do in git, for example.

- Bert -

smime.p7s (5K) Download Attachment

Chris Muller-3

Re: The Trunk: Chronology-Core-dtl.2.mcz

In reply to this post by Chris Muller-3

... and so I tried to load my personal preferences AFTER updating, but
the removal of InputSensor has rendered all .prefs files unloadable...

On Tue, Mar 1, 2016 at 2:15 PM, Chris Muller <[hidden email]> wrote:

>> I'm referring to the update stream being broken. A stock
>> Squeak5.0-15113 cannot currently be updated to latest trunk because of
>> that underscore. I wanted to make sure Dave wasn't confusing that
>> with his change.
>
> Correction: Okay, I didn't realize we allowUnderscoreAssignments by
> default. Really, still?
>
> When I make my trunk image, I have been loading my own prefs to set my
> updateUrl to trunk for myself, but it also includes setting this off,
> and so that's why it was blowing up for me.

Chris Muller-3

Re: The Trunk: Chronology-Core-dtl.2.mcz

In reply to this post by Bert Freudenberg

> And it looks like the differ replaces underscore assignments by ansi assignments. Maybe that’s why I missed one line when publishing my fix

diffsWithPrettyPrint