Smalltalk › Squeak › Squeak - Dev

The Trunk: PackageInfo-Base-nice.53.mcz

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

6 messages Options

commits-2

The Trunk: PackageInfo-Base-nice.53.mcz

Nicolas Cellier uploaded a new version of PackageInfo-Base to project The Trunk:
http://source.squeak.org/trunk/PackageInfo-Base-nice.53.mcz

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

Name: PackageInfo-Base-nice.53
Author: nice
Time: 16 February 2011, 12:02:09.324 am
UUID: 27c62178-0d9c-46ad-a2f2-a80b473740cb
Ancestors: PackageInfo-Base-cmm.52

A Metaclass now respondsTo: #category, so don't tell, just ask.

=============== Diff against PackageInfo-Base-cmm.52 ===============

Item was changed:
----- Method: PackageInfo>>includesClass: (in category 'testing') -----
includesClass: aClass
+ ^ self includesSystemCategory: aClass category!
- ^ self includesSystemCategory: aClass theNonMetaClass category!

Levente Uzonyi-2

Re: The Trunk: PackageInfo-Base-nice.53.mcz

On Tue, 15 Feb 2011, [hidden email] wrote:

> Nicolas Cellier uploaded a new version of PackageInfo-Base to project The Trunk:
> http://source.squeak.org/trunk/PackageInfo-Base-nice.53.mcz
>
> ==================== Summary ====================
>
> Name: PackageInfo-Base-nice.53
> Author: nice
> Time: 16 February 2011, 12:02:09.324 am
> UUID: 27c62178-0d9c-46ad-a2f2-a80b473740cb
> Ancestors: PackageInfo-Base-cmm.52
>
> A Metaclass now respondsTo: #category, so don't tell, just ask.

Isn't it necessary to create a new mcm? What if the update process will
try to use this method with a Metaclass, before Metaclass >> #category is
loaded? (I'm not sure it's a possible case, but it's safer to define the
load order IMHO.)

Levente

>
> =============== Diff against PackageInfo-Base-cmm.52 ===============
>
> Item was changed:
> ----- Method: PackageInfo>>includesClass: (in category 'testing') -----
> includesClass: aClass
> + ^ self includesSystemCategory: aClass category!
> - ^ self includesSystemCategory: aClass theNonMetaClass category!
>
>
>

Nicolas Cellier

Re: The Trunk: PackageInfo-Base-nice.53.mcz

2011/2/16 Levente Uzonyi <[hidden email]>:

> On Tue, 15 Feb 2011, [hidden email] wrote:
>
>> Nicolas Cellier uploaded a new version of PackageInfo-Base to project The
>> Trunk:
>> http://source.squeak.org/trunk/PackageInfo-Base-nice.53.mcz
>>
>> ==================== Summary ====================
>>
>> Name: PackageInfo-Base-nice.53
>> Author: nice
>> Time: 16 February 2011, 12:02:09.324 am
>> UUID: 27c62178-0d9c-46ad-a2f2-a80b473740cb
>> Ancestors: PackageInfo-Base-cmm.52
>>
>> A Metaclass now respondsTo: #category, so don't tell, just ask.
>
> Isn't it necessary to create a new mcm? What if the update process will try
> to use this method with a Metaclass, before Metaclass >> #category is
> loaded? (I'm not sure it's a possible case, but it's safer to define the
> load order IMHO.)
>
>
> Levente
>

As I understand it, if Kernel is before PackageInfo-Base in previous
mcm, then it's not required.
But sometimes, I have chicken skin when pressing the MC save button...

Nicolas

>>
>> =============== Diff against PackageInfo-Base-cmm.52 ===============
>>
>> Item was changed:
>> ----- Method: PackageInfo>>includesClass: (in category 'testing') -----
>> includesClass: aClass
>> + ^ self includesSystemCategory: aClass category!
>> - ^ self includesSystemCategory: aClass theNonMetaClass category!
>>
>>
>>
>
>

Levente Uzonyi-2

Re: The Trunk: PackageInfo-Base-nice.53.mcz

On Wed, 16 Feb 2011, Nicolas Cellier wrote:

> 2011/2/16 Levente Uzonyi <[hidden email]>:
>> On Tue, 15 Feb 2011, [hidden email] wrote:
>>
>>> Nicolas Cellier uploaded a new version of PackageInfo-Base to project The
>>> Trunk:
>>> http://source.squeak.org/trunk/PackageInfo-Base-nice.53.mcz
>>>
>>> ==================== Summary ====================
>>>
>>> Name: PackageInfo-Base-nice.53
>>> Author: nice
>>> Time: 16 February 2011, 12:02:09.324 am
>>> UUID: 27c62178-0d9c-46ad-a2f2-a80b473740cb
>>> Ancestors: PackageInfo-Base-cmm.52
>>>
>>> A Metaclass now respondsTo: #category, so don't tell, just ask.
>>
>> Isn't it necessary to create a new mcm? What if the update process will try
>> to use this method with a Metaclass, before Metaclass >> #category is
>> loaded? (I'm not sure it's a possible case, but it's safer to define the
>> load order IMHO.)
>>
>>
>> Levente
>>
>
> As I understand it, if Kernel is before PackageInfo-Base in previous
> mcm, then it's not required.

That's right. For some unknown reason I mixed PackageInfo-Base with
Monticello, sorry for that. Btw I used to create new mcms in cases like
this, just to be sure. :)

Levente

> But sometimes, I have chicken skin when pressing the MC save button...
>
> Nicolas
>
>>>
>>> =============== Diff against PackageInfo-Base-cmm.52 ===============
>>>
>>> Item was changed:
>>> ----- Method: PackageInfo>>includesClass: (in category 'testing') -----
>>> includesClass: aClass
>>> + ^ self includesSystemCategory: aClass category!
>>> - ^ self includesSystemCategory: aClass theNonMetaClass category!
>>>
>>>
>>>
>>
>>
>
>

Bert Freudenberg

MCMs (was Re: [squeak-dev] The Trunk: PackageInfo-Base-nice.53.mcz)

On 16.02.2011, at 02:06, Levente Uzonyi wrote:

> I used to create new mcms in cases like this, just to be sure. :)

Ah. And I wondered where all the MCMs come from. I had a hunch it's way too many ;)

You only need to create an MCM if you need to force loading a specific package version (and not a newer one), or if you need to temporarily change the load order.

- Bert -

Levente Uzonyi-2

Re: MCMs (was Re: [squeak-dev] The Trunk: PackageInfo-Base-nice.53.mcz)

On Wed, 16 Feb 2011, Bert Freudenberg wrote:

> On 16.02.2011, at 02:06, Levente Uzonyi wrote:
>
>> I used to create new mcms in cases like this, just to be sure. :)
>
> Ah. And I wondered where all the MCMs come from. I had a hunch it's way too many ;)

Most of my mcms are not like this. Usually the load order has to be
defined for the changes of a single package.

>
> You only need to create an MCM if you need to force loading a specific package version (and not a newer one), or if you need to temporarily change the load order.

This is how I created mcms so far, though IIRC I only changed the order of
the packages within an mcm only once. It's easier to create an extra mcm
than reordering the package list. :)

Levente

>
> - Bert -
>
>
>
>