On Sun, Mar 24, 2013 at 05:33:13AM +0000, commits@source.squeak.org wrote:
tim Rowledge uploaded a new version of Graphics to project The Trunk: http://source.squeak.org/trunk/Graphics-tpr.205.mcz
==================== Summary ====================
Name: Graphics-tpr.205 Author: tpr Time: 23 March 2013, 10:32:54.898 pm UUID: a16cf25b-36b7-4a5e-b3e9-3786b5f7c92d Ancestors: Graphics-tpr.204
A somewhat involved pile of changes that I'm not 10% sure my use of MC has correctly caught; revert this if you find bugs! For example, MC claims that AbstractFont is very different but visual inspection says not.
Indeed, it looks like this update removes a number of AbstractFont methods. I tried loading it and it hangs my image.
Dave
On 24-03-2013, at 9:29 AM, "David T. Lewis" lewis@mail.msen.com wrote:
On Sun, Mar 24, 2013 at 05:33:13AM +0000, commits@source.squeak.org wrote:
tim Rowledge uploaded a new version of Graphics to project The Trunk: http://source.squeak.org/trunk/Graphics-tpr.205.mcz
==================== Summary ====================
Name: Graphics-tpr.205 Author: tpr Time: 23 March 2013, 10:32:54.898 pm UUID: a16cf25b-36b7-4a5e-b3e9-3786b5f7c92d Ancestors: Graphics-tpr.204
A somewhat involved pile of changes that I'm not 10% sure my use of MC has correctly caught; revert this if you find bugs! For example, MC claims that AbstractFont is very different but visual inspection says not.
Indeed, it looks like this update removes a number of AbstractFont methods. I tried loading it and it hangs my image.
Damn, that's what I was afraid of.
Thing is, I merged the relevant prior version, added my changes that have nothing at all to do with AbstractFont and saved the end result. At least that leaves me feeling that it wasn't really my fault…
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Strange OpCodes: SDR: Shift Disk Right
On Sun, Mar 24, 2013 at 10:31:08AM -0700, tim Rowledge wrote:
On 24-03-2013, at 9:29 AM, "David T. Lewis" lewis@mail.msen.com wrote:
On Sun, Mar 24, 2013 at 05:33:13AM +0000, commits@source.squeak.org wrote:
tim Rowledge uploaded a new version of Graphics to project The Trunk: http://source.squeak.org/trunk/Graphics-tpr.205.mcz
==================== Summary ====================
Name: Graphics-tpr.205 Author: tpr Time: 23 March 2013, 10:32:54.898 pm UUID: a16cf25b-36b7-4a5e-b3e9-3786b5f7c92d Ancestors: Graphics-tpr.204
A somewhat involved pile of changes that I'm not 10% sure my use of MC has correctly caught; revert this if you find bugs! For example, MC claims that AbstractFont is very different but visual inspection says not.
Indeed, it looks like this update removes a number of AbstractFont methods. I tried loading it and it hangs my image.
Damn, that's what I was afraid of.
Thing is, I merged the relevant prior version, added my changes that have nothing at all to do with AbstractFont and saved the end result. At least that leaves me feeling that it wasn't really my fault?
I'm not sure what the problem is. I was confused by the "differences" I was seeing immediately prior to applying your update. Unfortunately I'm heading out the door right now and can't follow up on anything, so I didn't want to try to fix something and make it worse. Maybe someone else can have a look and suggest what to do?
Dave
Most probably you moved AbstractFont to another package. Could you verify how it is classified? I have :
Object subclass: #AbstractFont instanceVariableNames: '' classVariableNames: 'ForceNonSubPixelCount' poolDictionaries: '' ---> category: 'Graphics-Fonts'
If it's not the case, change the category and publish a new version of Graphics.
Also the fact that you published 'nil' instead of an empty postscript maked me think you don't have an up to date image, because I think Bert solved this one already.
+ (PackageInfo named: 'Graphics') postscriptOfRemoval: 'nil'!
Nicolas
2013/3/24 David T. Lewis lewis@mail.msen.com:
On Sun, Mar 24, 2013 at 10:31:08AM -0700, tim Rowledge wrote:
On 24-03-2013, at 9:29 AM, "David T. Lewis" lewis@mail.msen.com wrote:
On Sun, Mar 24, 2013 at 05:33:13AM +0000, commits@source.squeak.org wrote:
tim Rowledge uploaded a new version of Graphics to project The Trunk: http://source.squeak.org/trunk/Graphics-tpr.205.mcz
==================== Summary ====================
Name: Graphics-tpr.205 Author: tpr Time: 23 March 2013, 10:32:54.898 pm UUID: a16cf25b-36b7-4a5e-b3e9-3786b5f7c92d Ancestors: Graphics-tpr.204
A somewhat involved pile of changes that I'm not 10% sure my use of MC has correctly caught; revert this if you find bugs! For example, MC claims that AbstractFont is very different but visual inspection says not.
Indeed, it looks like this update removes a number of AbstractFont methods. I tried loading it and it hangs my image.
Damn, that's what I was afraid of.
Thing is, I merged the relevant prior version, added my changes that have nothing at all to do with AbstractFont and saved the end result. At least that leaves me feeling that it wasn't really my fault?
I'm not sure what the problem is. I was confused by the "differences" I was seeing immediately prior to applying your update. Unfortunately I'm heading out the door right now and can't follow up on anything, so I didn't want to try to fix something and make it worse. Maybe someone else can have a look and suggest what to do?
Dave
2013/3/24 Nicolas Cellier nicolas.cellier.aka.nice@gmail.com:
Most probably you moved AbstractFont to another package. Could you verify how it is classified? I have :
Object subclass: #AbstractFont instanceVariableNames: '' classVariableNames: 'ForceNonSubPixelCount' poolDictionaries: '' ---> category: 'Graphics-Fonts'
If it's not the case, change the category and publish a new version of Graphics.
Also the fact that you published 'nil' instead of an empty postscript maked me think you don't have an up to date image, because I think
^H^H^H^H^Hmakes
Bert solved this one already.
- (PackageInfo named: 'Graphics') postscriptOfRemoval: 'nil'!
Nicolas
2013/3/24 David T. Lewis lewis@mail.msen.com:
On Sun, Mar 24, 2013 at 10:31:08AM -0700, tim Rowledge wrote:
On 24-03-2013, at 9:29 AM, "David T. Lewis" lewis@mail.msen.com wrote:
On Sun, Mar 24, 2013 at 05:33:13AM +0000, commits@source.squeak.org wrote:
tim Rowledge uploaded a new version of Graphics to project The Trunk: http://source.squeak.org/trunk/Graphics-tpr.205.mcz
==================== Summary ====================
Name: Graphics-tpr.205 Author: tpr Time: 23 March 2013, 10:32:54.898 pm UUID: a16cf25b-36b7-4a5e-b3e9-3786b5f7c92d Ancestors: Graphics-tpr.204
A somewhat involved pile of changes that I'm not 10% sure my use of MC has correctly caught; revert this if you find bugs! For example, MC claims that AbstractFont is very different but visual inspection says not.
Indeed, it looks like this update removes a number of AbstractFont methods. I tried loading it and it hangs my image.
Damn, that's what I was afraid of.
Thing is, I merged the relevant prior version, added my changes that have nothing at all to do with AbstractFont and saved the end result. At least that leaves me feeling that it wasn't really my fault?
I'm not sure what the problem is. I was confused by the "differences" I was seeing immediately prior to applying your update. Unfortunately I'm heading out the door right now and can't follow up on anything, so I didn't want to try to fix something and make it worse. Maybe someone else can have a look and suggest what to do?
Dave
On 24-03-2013, at 11:04 AM, Nicolas Cellier nicolas.cellier.aka.nice@gmail.com wrote:
Most probably you moved AbstractFont to another package. Could you verify how it is classified? I have :
Object subclass: #AbstractFont instanceVariableNames: '' classVariableNames: 'ForceNonSubPixelCount' poolDictionaries: '' ---> category: 'Graphics-Fonts'
Not me; it must be an artefact of the image I was using, which it turns out to my surprise is a 4.3 version.
I've been trying to get a 4.4-12337 and update packages but so far it mostly seems to blow up. Some issue with SketchMorph and #rotationCenter: is one problem and trying to load the latest compiler package just trashes the vm (the Cog vm, so far as I can tell from the dump file 'CogVM VM 6.0-pre - 21.0')
I've attempted to find a newer image than the 4.4 but nothing seems to exist. While looking for a newer Mac Cog vm http://www.squeakvm.org/mac/ points to a 404. http://www.mirandabanda.org/files/Cog/VM/VM.r2701/cog.app.tgz appears to be about right though it disconcertingly claims to be Croquet instead of Squeak - no idea if that has any real meaning.
OK, so next I try using the latest 'update' entry in the trunk repository. Loading that a) complains about loading 'Exceptions' on top of a package that has unsaved changes in the image. That doesn't sound good. Why does a released image have uncommitted code changes? b) MNU for #undeclared in SystemDictionary while loading system-fbs.519 Also not good sounding. It appears that the only implementor of #undeclared is Environment, but the environment iVar in Class>declare is still the SystemDictionary. Boom. It may be of interest that the still extant 'Undeclared' dictionary contains #EnvironmentRequest (nil), #category (#'public access'), #class (Compiler), #context (nil) and #requestor (nil).
I'm not where to go from here. It seems a bit alarming that it is so problematic to update an image before I can even try to submit the changes I'm actually interested in.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Concurrent: an object that looks like a raisin but isn't
That's many rants, I don't think I can answer to all of them. An up to date image is built automatically:
http://build.squeak.org/job/SqueakTrunk
If you browse the site, you might also get an unofficial VM automatically built.
Updating a modified image has no guaranty to work when one modifies the system too deeply... And just loading the latest MC version has very few chances to work, some intermediate stages are mandatory. In the most complex cases, one must merge the modifications of the older image from an up-to-date image.
Last thing, I think that the dirty Exceptions package is not really dirty. Most probably it's due to an incorrect load order in a mcm.
Nicolas
2013/3/24 tim Rowledge tim@rowledge.org:
On 24-03-2013, at 11:04 AM, Nicolas Cellier nicolas.cellier.aka.nice@gmail.com wrote:
Most probably you moved AbstractFont to another package. Could you verify how it is classified? I have :
Object subclass: #AbstractFont instanceVariableNames: '' classVariableNames: 'ForceNonSubPixelCount' poolDictionaries: '' ---> category: 'Graphics-Fonts'
Not me; it must be an artefact of the image I was using, which it turns out to my surprise is a 4.3 version.
I've been trying to get a 4.4-12337 and update packages but so far it mostly seems to blow up. Some issue with SketchMorph and #rotationCenter: is one problem and trying to load the latest compiler package just trashes the vm (the Cog vm, so far as I can tell from the dump file 'CogVM VM 6.0-pre - 21.0')
I've attempted to find a newer image than the 4.4 but nothing seems to exist. While looking for a newer Mac Cog vm http://www.squeakvm.org/mac/ points to a 404. http://www.mirandabanda.org/files/Cog/VM/VM.r2701/cog.app.tgz appears to be about right though it disconcertingly claims to be Croquet instead of Squeak - no idea if that has any real meaning.
OK, so next I try using the latest 'update' entry in the trunk repository. Loading that a) complains about loading 'Exceptions' on top of a package that has unsaved changes in the image. That doesn't sound good. Why does a released image have uncommitted code changes? b) MNU for #undeclared in SystemDictionary while loading system-fbs.519 Also not good sounding. It appears that the only implementor of #undeclared is Environment, but the environment iVar in Class>declare is still the SystemDictionary. Boom. It may be of interest that the still extant 'Undeclared' dictionary contains #EnvironmentRequest (nil), #category (#'public access'), #class (Compiler), #context (nil) and #requestor (nil).
I'm not where to go from here. It seems a bit alarming that it is so problematic to update an image before I can even try to submit the changes I'm actually interested in.
tim
tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Concurrent: an object that looks like a raisin but isn't
On 24-03-2013, at 1:42 PM, Nicolas Cellier nicolas.cellier.aka.nice@gmail.com wrote:
An up to date image is built automatically:
So it is. Now, wouldn't it be a nice idea if the main squeak.org download page pointed to that, along with a brief explanation of what to expect?
If you browse the site, you might also get an unofficial VM automatically built.
Updating a modified image has no guaranty to work when one modifies the system too deeply…
Although my initial problems were in a lightly altered image, the latter issues were with a fresh, untouched, virgin official release image...
And just loading the latest MC version has very few chances to work, some intermediate stages are mandatory.
I claim that loading an official configuration from the trunk into a clean official image really ought not have any problems.
Last thing, I think that the dirty Exceptions package is not really dirty. Most probably it's due to an incorrect load order in a mcm.
Sounds like a bug to me.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Useful random insult:- In serious need of attitude adjustment.
On 24-03-2013, at 1:42 PM, Nicolas Cellier nicolas.cellier.aka.nice@gmail.com wrote:
Last thing, I think that the dirty Exceptions package is not really dirty. Most probably it's due to an incorrect load order in a mcm.
Running the Trunk image you pointed to I see that the Exceptions package claim to be dirty but after asking to see the changes from the trunk repository it declares itself clean. This seems like something that would be trivial to clear up given the appropriate file, permissions and path to save it to.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Strange OpCodes: PSP: Push Stack Pointer
On 24-03-2013, at 2:36 PM, tim Rowledge tim@rowledge.org wrote:
On 24-03-2013, at 1:42 PM, Nicolas Cellier nicolas.cellier.aka.nice@gmail.com wrote:
Last thing, I think that the dirty Exceptions package is not really dirty. Most probably it's due to an incorrect load order in a mcm.
Running the Trunk image you pointed to I see that the Exceptions package claim to be dirty but after asking to see the changes from the trunk repository it declares itself clean. This seems like something that would be trivial to clear up given the appropriate file, permissions and path to save it to.
Likewise 'System'
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim taffeta - Welsh goats cheese
On 24 March 2013 21:37, tim Rowledge tim@rowledge.org wrote:
On 24-03-2013, at 2:36 PM, tim Rowledge tim@rowledge.org wrote:
On 24-03-2013, at 1:42 PM, Nicolas Cellier nicolas.cellier.aka.nice@gmail.com wrote:
Last thing, I think that the dirty Exceptions package is not really dirty. Most probably it's due to an incorrect load order in a mcm.
Running the Trunk image you pointed to I see that the Exceptions package claim to be dirty but after asking to see the changes from the trunk repository it declares itself clean. This seems like something that would be trivial to clear up given the appropriate file, permissions and path to save it to.
Likewise 'System'
The packages should not be marked as dirty. The release process in fact is supposed to protect exactly this. I've recorded the bug here: - http://bugs.squeak.org/view.php?id=7759 - and hopefully I'll be able to take a look at it soon.
frank
tim
tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim taffeta - Welsh goats cheese
On 24 March 2013 23:01, Frank Shearar frank.shearar@gmail.com wrote:
On 24 March 2013 21:37, tim Rowledge tim@rowledge.org wrote:
On 24-03-2013, at 2:36 PM, tim Rowledge tim@rowledge.org wrote:
On 24-03-2013, at 1:42 PM, Nicolas Cellier nicolas.cellier.aka.nice@gmail.com wrote:
Last thing, I think that the dirty Exceptions package is not really dirty. Most probably it's due to an incorrect load order in a mcm.
Running the Trunk image you pointed to I see that the Exceptions package claim to be dirty but after asking to see the changes from the trunk repository it declares itself clean. This seems like something that would be trivial to clear up given the appropriate file, permissions and path to save it to.
Likewise 'System'
The packages should not be marked as dirty. The release process in fact is supposed to protect exactly this. I've recorded the bug here:
- http://bugs.squeak.org/view.php?id=7759 - and hopefully I'll be able
to take a look at it soon.
I don't know the exact provenance of Tim's package, but the product of ReleaseSqueakTrunk should be a pristine image. Certainly this guy is: http://build.squeak.org/job/ReleaseSqueakTrunk/146/artifact/target/Squeak4.5...
I've closed Mantis 7759.
frank
frank
tim
tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim taffeta - Welsh goats cheese
On 26-03-2013, at 2:04 PM, Frank Shearar frank.shearar@gmail.com wrote:
I don't know the exact provenance of Tim's package, but the product of ReleaseSqueakTrunk should be a pristine image. Certainly this guy is: http://build.squeak.org/job/ReleaseSqueakTrunk/146/artifact/target/Squeak4.5...
It was (According to cmd-i on my iMac) http://build.squeak.org/job/SqueakTrunk/lastSuccessfulBuild/artifact/target/... as of the 24th.
The image was *actually* clean as I mentioned but MC marked Exceptions and System as dirty until I checked them, at which point it decided they were in fact clean.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Useful random insult:- Kept an open mind -- and his brains fell out.
On 26 March 2013 21:56, tim Rowledge tim@rowledge.org wrote:
On 26-03-2013, at 2:04 PM, Frank Shearar frank.shearar@gmail.com wrote:
I don't know the exact provenance of Tim's package, but the product of ReleaseSqueakTrunk should be a pristine image. Certainly this guy is: http://build.squeak.org/job/ReleaseSqueakTrunk/146/artifact/target/Squeak4.5...
It was (According to cmd-i on my iMac) http://build.squeak.org/job/SqueakTrunk/lastSuccessfulBuild/artifact/target/... as of the 24th.
The image was *actually* clean as I mentioned but MC marked Exceptions and System as dirty until I checked them, at which point it decided they were in fact clean.
OK. TrunkImage artifacts _should_ be clean, but I make no guarantees. It's only ReleaseSqueakTrunk that runs all the check-packages-are-clean type processes.
Clearly it's not obvious that ReleaseSqueakTrunk is the proper "public" artifact maker: artifacts that it makes ought to be of such a state that they may be pushed to ftp.squeak.org.
TrunkImage, in contrast, takes a base image and updates it, but does nothing special.
At various points in the future TrunkImage will have unloadable packages ripped out (to be reloaded by ReleaseSqueakTrunk).
Where might be a good place to document these artifacts and artifact-producers?
frank
squeak-dev@lists.squeakfoundation.org
-
David T. Lewis -
Frank Shearar -
Nicolas Cellier -
tim Rowledge