On Tue, 15 Feb 2011, commits@source.squeak.org 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!
2011/2/16 Levente Uzonyi leves@elte.hu:
On Tue, 15 Feb 2011, commits@source.squeak.org 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!
On Wed, 16 Feb 2011, Nicolas Cellier wrote:
2011/2/16 Levente Uzonyi leves@elte.hu:
On Tue, 15 Feb 2011, commits@source.squeak.org 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!
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 -
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 -
squeak-dev@lists.squeakfoundation.org
-
Bert Freudenberg -
Levente Uzonyi -
Nicolas Cellier