Squeak4.2-10856-beta.zip
Locked
12
12
 
|
Hi all, I heard back from Andreas, he is fine with pushing the
FloatMathPlugin enhancements to 4.3. I loaded my latest version of KernelTests from the inbox which removes FloatConsistencyTests, and have just uploaded a first 4.2 beta to ftp://ftp.squeak.org/4.2/ We can try to go-back in the trunk stream so that a continuous update could work from a 10779 image, but I'm really not sure what needs to be done to accomplish that or whether that's important. We have this beta image from which we can at least update forward. We do still have some issues in the test-suite that we need help with! Please also test your applications with this image over the next weeks to ensure they will be compatible with this new release, and also flush out any final problems. We're getting there, thanks to all for your help. - Chris |
Re: Squeak4.2-10856-beta.zip
|
|
On Thu, 6 Jan 2011, Chris Muller wrote:
> Hi all, I heard back from Andreas, he is fine with pushing the > FloatMathPlugin enhancements to 4.3. > > I loaded my latest version of KernelTests from the inbox which removes > FloatConsistencyTests, and have just uploaded a first 4.2 beta to IMHO the release has to be a Trunk image. Manual changes must not change the code in the image, because that breaks the update mechanism. > > ftp://ftp.squeak.org/4.2/ > > We can try to go-back in the trunk stream so that a continuous update > could work from a 10779 image, but I'm really not sure what needs to > be done to accomplish that or whether that's important. We have this > beta image from which we can at least update forward. What would be the benefit from rejecting all changes since 10779 from 4.2? > > We do still have some issues in the test-suite that we need help with! AFAIK all but one failing tests have fixes in the Inbox. Levente > > Please also test your applications with this image over the next weeks > to ensure they will be compatible with this new release, and also > flush out any final problems. > > We're getting there, thanks to all for your help. > > - Chris > > |
Re: Squeak4.2-10856-beta.zip
|
|
In reply to this post by Chris Muller-4
Since the inception of Trunk, we've rolled a new image more or less whenever updating the last one got to be painful.
I think it would be great to apply the updates to the 4.1 release image and verify that it can be made current that way though. It's totally a job for Hudson, but I should be able to do that (time permitting) manually tonight and report on my results if there's interest from the community. On Jan 6, 2011, at 9:33 AM, Chris Muller <[hidden email]> wrote: > Hi all, I heard back from Andreas, he is fine with pushing the > FloatMathPlugin enhancements to 4.3. > > I loaded my latest version of KernelTests from the inbox which removes > FloatConsistencyTests, and have just uploaded a first 4.2 beta to > > ftp://ftp.squeak.org/4.2/ > > We can try to go-back in the trunk stream so that a continuous update > could work from a 10779 image, but I'm really not sure what needs to > be done to accomplish that or whether that's important. We have this > beta image from which we can at least update forward. > > We do still have some issues in the test-suite that we need help with! > > Please also test your applications with this image over the next weeks > to ensure they will be compatible with this new release, and also > flush out any final problems. > > We're getting there, thanks to all for your help. > > - Chris > |
|
|
In reply to this post by Chris Muller-4
Hi Chris,
Thanks for the 4.2 beta image. Am 06.01.2011 um 18:33 schrieb Chris Muller: > We can try to go-back in the trunk stream so that a continuous update > could work from a 10779 image, but I'm really not sure what needs to > be done to accomplish that or whether that's important. We have this > beta image from which we can at least update forward. I don't know if I understand you correctly, but IMHO it is important to keep the updates working so that people can update their development images to 4.2 without having start from a new image. > We do still have some issues in the test-suite that we need help with! Well, I tried. ;-) Cheers, Bernhard |
|
|
In reply to this post by Chris Muller-4
Am 06.01.2011 um 18:33 schrieb Chris Muller:
> Please also test your applications with this image over the next weeks > to ensure they will be compatible with this new release, and also > flush out any final problems. I think there is something wrong with the PackageDependencyTest in the beta image. I did not understand why testSunitUI does not fail without my fixes and looked into it. All 35 tests pass but it runs very fast. In the trunk image it takes several seconds. When I do DependencyBrowser open it is completely empty. Cheers, Bernhard |
|
|
In reply to this post by Levente Uzonyi-2
Am 06.01.2011 um 19:17 schrieb Levente Uzonyi:
> On Thu, 6 Jan 2011, Chris Muller wrote: > >> Hi all, I heard back from Andreas, he is fine with pushing the >> FloatMathPlugin enhancements to 4.3. >> >> I loaded my latest version of KernelTests from the inbox which removes >> FloatConsistencyTests, and have just uploaded a first 4.2 beta to > > IMHO the release has to be a Trunk image. Manual changes must not change the code in the image, because that breaks the update mechanism. I agree that this would be a very good thing. Just my 2 cents. Cheers, Bernhard |
|
|
In reply to this post by Chris Muller-4
One comment: The "Welcome to Squeak 4.2" workspace is still all 4.1
content. I think it would be best to simply copy the "What's New" section into that workspace. Cheers, - Andreas On 1/6/2011 6:33 PM, Chris Muller wrote: > Hi all, I heard back from Andreas, he is fine with pushing the > FloatMathPlugin enhancements to 4.3. > > I loaded my latest version of KernelTests from the inbox which removes > FloatConsistencyTests, and have just uploaded a first 4.2 beta to > > ftp://ftp.squeak.org/4.2/ > > We can try to go-back in the trunk stream so that a continuous update > could work from a 10779 image, but I'm really not sure what needs to > be done to accomplish that or whether that's important. We have this > beta image from which we can at least update forward. > > We do still have some issues in the test-suite that we need help with! > > Please also test your applications with this image over the next weeks > to ensure they will be compatible with this new release, and also > flush out any final problems. > > We're getting there, thanks to all for your help. > > - Chris > > |
|
|
In reply to this post by Chris Muller-4
Hi Chris,
I saw that you set the #swapMouseButtons preference. Now the right mouse button is fixed on the Mac. However, I just tested the image under Windows. I am afraid now it is broken there, i.e. the right mouse shows the halos instead of the context menu. There have been numerous threads about this problem. IMHO the right solution would be: - Make the release VMs consistent so that one can move an image between the platforms and the right mouse button behavior stays the same without need to change preferences (or run platform specific code like Pharo does it IIRC). - As Mac and Linux seem to behave the same the easiest way seem to me to change the Squeak*.ini to include 3ButtonMouse=1. I fully agree with Igor on that topic: http://lists.squeakfoundation.org/pipermail/squeak-dev/2010-December/155896.html Cheers, Bernhard Am 06.01.2011 um 18:33 schrieb Chris Muller: > Hi all, I heard back from Andreas, he is fine with pushing the > FloatMathPlugin enhancements to 4.3. > > I loaded my latest version of KernelTests from the inbox which removes > FloatConsistencyTests, and have just uploaded a first 4.2 beta to > > ftp://ftp.squeak.org/4.2/ > > We can try to go-back in the trunk stream so that a continuous update > could work from a 10779 image, but I'm really not sure what needs to > be done to accomplish that or whether that's important. We have this > beta image from which we can at least update forward. > > We do still have some issues in the test-suite that we need help with! > > Please also test your applications with this image over the next weeks > to ensure they will be compatible with this new release, and also > flush out any final problems. > > We're getting there, thanks to all for your help. > > - Chris > |
Re: Right mouse button again (Was Re: Squeak4.2-10856-beta.zip)
|
|
Please everyone, let's not make this any more complex than it absolutely
needs to be. Here are the guidelines: a) The behavior of the Windows VM *changes* between 4.1 and 4.2 and this change will be documented. b) As a consequence, if you use a 4.2 beta image with a 4.1 VM you will need to adjust the mouse button preference manually. c) The 4.2 release will include VMs with the preference set "correctly". That's it. In other words, there is nothing "broken" at this point, the behavior you are seeing is entirely expected because of a). For the records, adding more platform specific code is the *wrong* thing to do and just adds useless bloat. We will go through a short period of pain for people who use 4.2 beta with 4.1 VMs but that's what happens when you change things. It'll be over in a few weeks. Cheers, - Andreas On 1/6/2011 9:46 PM, Bernhard Pieber wrote: > Hi Chris, > > I saw that you set the #swapMouseButtons preference. Now the right mouse button is fixed on the Mac. However, I just tested the image under Windows. I am afraid now it is broken there, i.e. the right mouse shows the halos instead of the context menu. > > There have been numerous threads about this problem. IMHO the right solution would be: > - Make the release VMs consistent so that one can move an image between the platforms and the right mouse button behavior stays the same without need to change preferences (or run platform specific code like Pharo does it IIRC). > - As Mac and Linux seem to behave the same the easiest way seem to me to change the Squeak*.ini to include 3ButtonMouse=1. > > I fully agree with Igor on that topic: > http://lists.squeakfoundation.org/pipermail/squeak-dev/2010-December/155896.html > > Cheers, > Bernhard > > Am 06.01.2011 um 18:33 schrieb Chris Muller: > >> Hi all, I heard back from Andreas, he is fine with pushing the >> FloatMathPlugin enhancements to 4.3. >> >> I loaded my latest version of KernelTests from the inbox which removes >> FloatConsistencyTests, and have just uploaded a first 4.2 beta to >> >> ftp://ftp.squeak.org/4.2/ >> >> We can try to go-back in the trunk stream so that a continuous update >> could work from a 10779 image, but I'm really not sure what needs to >> be done to accomplish that or whether that's important. We have this >> beta image from which we can at least update forward. >> >> We do still have some issues in the test-suite that we need help with! >> >> Please also test your applications with this image over the next weeks >> to ensure they will be compatible with this new release, and also >> flush out any final problems. >> >> We're getting there, thanks to all for your help. >> >> - Chris >> > > > |
|
|
Am 06.01.2011 um 22:04 schrieb Andreas Raab:
> Please everyone, let's not make this any more complex than it absolutely needs to be. Here are the guidelines: > a) The behavior of the Windows VM *changes* between 4.1 and 4.2 and this change will be documented. Sorry, I missed that this was already decided. This is great! I did not want to make it more complex. > b) As a consequence, if you use a 4.2 beta image with a 4.1 VM you will need to adjust the mouse button preference manually. > c) The 4.2 release will include VMs with the preference set "correctly". > > That's it. In other words, there is nothing "broken" at this point, the behavior you are seeing is entirely expected because of a). For the records, adding more platform specific code is the *wrong* thing to do and just adds useless bloat. This is great as well! > We will go through a short period of pain for people who use 4.2 beta with 4.1 VMs but that's what happens when you change things. It'll be over in a few weeks. > > Cheers, > - Andreas > > On 1/6/2011 9:46 PM, Bernhard Pieber wrote: >> Hi Chris, >> >> I saw that you set the #swapMouseButtons preference. Now the right mouse button is fixed on the Mac. However, I just tested the image under Windows. I am afraid now it is broken there, i.e. the right mouse shows the halos instead of the context menu. >> >> There have been numerous threads about this problem. IMHO the right solution would be: >> - Make the release VMs consistent so that one can move an image between the platforms and the right mouse button behavior stays the same without need to change preferences (or run platform specific code like Pharo does it IIRC). >> - As Mac and Linux seem to behave the same the easiest way seem to me to change the Squeak*.ini to include 3ButtonMouse=1. >> >> I fully agree with Igor on that topic: >> http://lists.squeakfoundation.org/pipermail/squeak-dev/2010-December/155896.html >> >> Cheers, >> Bernhard |
Re: Right mouse button again (Was Re: Squeak4.2-10856-beta.zip)
|
|
On 1/6/2011 10:24 PM, Bernhard Pieber wrote:
> Am 06.01.2011 um 22:04 schrieb Andreas Raab: >> Please everyone, let's not make this any more complex than it absolutely needs to be. Here are the guidelines: >> a) The behavior of the Windows VM *changes* between 4.1 and 4.2 and this change will be documented. > Sorry, I missed that this was already decided. This is great! I did not want to make it more complex. Yes, I know :-) I just want to make sure people understand that the current state is both expected as well as temporary. Cheers, - Andreas >> b) As a consequence, if you use a 4.2 beta image with a 4.1 VM you will need to adjust the mouse button preference manually. >> c) The 4.2 release will include VMs with the preference set "correctly". >> >> That's it. In other words, there is nothing "broken" at this point, the behavior you are seeing is entirely expected because of a). For the records, adding more platform specific code is the *wrong* thing to do and just adds useless bloat. > This is great as well! >> We will go through a short period of pain for people who use 4.2 beta with 4.1 VMs but that's what happens when you change things. It'll be over in a few weeks. >> >> Cheers, >> - Andreas >> >> On 1/6/2011 9:46 PM, Bernhard Pieber wrote: >>> Hi Chris, >>> >>> I saw that you set the #swapMouseButtons preference. Now the right mouse button is fixed on the Mac. However, I just tested the image under Windows. I am afraid now it is broken there, i.e. the right mouse shows the halos instead of the context menu. >>> >>> There have been numerous threads about this problem. IMHO the right solution would be: >>> - Make the release VMs consistent so that one can move an image between the platforms and the right mouse button behavior stays the same without need to change preferences (or run platform specific code like Pharo does it IIRC). >>> - As Mac and Linux seem to behave the same the easiest way seem to me to change the Squeak*.ini to include 3ButtonMouse=1. >>> >>> I fully agree with Igor on that topic: >>> http://lists.squeakfoundation.org/pipermail/squeak-dev/2010-December/155896.html >>> >>> Cheers, >>> Bernhard > > |
|
|
In reply to this post by Andreas.Raab
Good catch. Am I reading an implicit suggestion in your note to close
the window too, so that Squeak 4.2 launches with an empty desktop? I could appreciate that "clean" look. What is your preference about that? On Thu, Jan 6, 2011 at 2:30 PM, Andreas Raab <[hidden email]> wrote: > One comment: The "Welcome to Squeak 4.2" workspace is still all 4.1 content. > I think it would be best to simply copy the "What's New" section into that > workspace. > > Cheers, > - Andreas > > On 1/6/2011 6:33 PM, Chris Muller wrote: >> >> Hi all, I heard back from Andreas, he is fine with pushing the >> FloatMathPlugin enhancements to 4.3. >> >> I loaded my latest version of KernelTests from the inbox which removes >> FloatConsistencyTests, and have just uploaded a first 4.2 beta to >> >> ftp://ftp.squeak.org/4.2/ >> >> We can try to go-back in the trunk stream so that a continuous update >> could work from a 10779 image, but I'm really not sure what needs to >> be done to accomplish that or whether that's important. We have this >> beta image from which we can at least update forward. >> >> We do still have some issues in the test-suite that we need help with! >> >> Please also test your applications with this image over the next weeks >> to ensure they will be compatible with this new release, and also >> flush out any final problems. >> >> We're getting there, thanks to all for your help. >> >> - Chris >> >> > > > |
Re: Right mouse button again (Was Re: [squeak-dev] Squeak4.2-10856-beta.zip)
|
|
In reply to this post by bpi
Hi, yes, I put that in there because there was a test added by Matthew
Fulmer that failed if it wasn't set. I think I removed the test and put it into ReleaseBuilderTrunk instead. It's perfectly fine with me to turn that change back. Ok with you, Matthew? - Chris On Thu, Jan 6, 2011 at 2:46 PM, Bernhard Pieber <[hidden email]> wrote: > Hi Chris, > > I saw that you set the #swapMouseButtons preference. Now the right mouse button is fixed on the Mac. However, I just tested the image under Windows. I am afraid now it is broken there, i.e. the right mouse shows the halos instead of the context menu. > > There have been numerous threads about this problem. IMHO the right solution would be: > - Make the release VMs consistent so that one can move an image between the platforms and the right mouse button behavior stays the same without need to change preferences (or run platform specific code like Pharo does it IIRC). > - As Mac and Linux seem to behave the same the easiest way seem to me to change the Squeak*.ini to include 3ButtonMouse=1. > > I fully agree with Igor on that topic: > http://lists.squeakfoundation.org/pipermail/squeak-dev/2010-December/155896.html > > Cheers, > Bernhard > > Am 06.01.2011 um 18:33 schrieb Chris Muller: > >> Hi all, I heard back from Andreas, he is fine with pushing the >> FloatMathPlugin enhancements to 4.3. >> >> I loaded my latest version of KernelTests from the inbox which removes >> FloatConsistencyTests, and have just uploaded a first 4.2 beta to >> >> ftp://ftp.squeak.org/4.2/ >> >> We can try to go-back in the trunk stream so that a continuous update >> could work from a 10779 image, but I'm really not sure what needs to >> be done to accomplish that or whether that's important. We have this >> beta image from which we can at least update forward. >> >> We do still have some issues in the test-suite that we need help with! >> >> Please also test your applications with this image over the next weeks >> to ensure they will be compatible with this new release, and also >> flush out any final problems. >> >> We're getting there, thanks to all for your help. >> >> - Chris >> > > > |
Re: Squeak4.2-10856-beta.zip
|
|
In reply to this post by Chris Muller-3
I like a welcome workspace. More than one is annoying to me though. That's just me though. I can see the "clean desktop" argument, and would probably prefer it, if I didn't feel that we were experiencing community growth. I think the welcome workspace is friendly to new folks.
On Jan 6, 2011, at 2:28 PM, Chris Muller <[hidden email]> wrote: > Good catch. Am I reading an implicit suggestion in your note to close > the window too, so that Squeak 4.2 launches with an empty desktop? I > could appreciate that "clean" look. What is your preference about > that? > > > On Thu, Jan 6, 2011 at 2:30 PM, Andreas Raab <[hidden email]> wrote: >> One comment: The "Welcome to Squeak 4.2" workspace is still all 4.1 content. >> I think it would be best to simply copy the "What's New" section into that >> workspace. >> >> Cheers, >> - Andreas >> >> On 1/6/2011 6:33 PM, Chris Muller wrote: >>> >>> Hi all, I heard back from Andreas, he is fine with pushing the >>> FloatMathPlugin enhancements to 4.3. >>> >>> I loaded my latest version of KernelTests from the inbox which removes >>> FloatConsistencyTests, and have just uploaded a first 4.2 beta to >>> >>> ftp://ftp.squeak.org/4.2/ >>> >>> We can try to go-back in the trunk stream so that a continuous update >>> could work from a 10779 image, but I'm really not sure what needs to >>> be done to accomplish that or whether that's important. We have this >>> beta image from which we can at least update forward. >>> >>> We do still have some issues in the test-suite that we need help with! >>> >>> Please also test your applications with this image over the next weeks >>> to ensure they will be compatible with this new release, and also >>> flush out any final problems. >>> >>> We're getting there, thanks to all for your help. >>> >>> - Chris >>> >>> >> >> >> > |
Re: Right mouse button again (Was Re: [squeak-dev] Squeak4.2-10856-beta.zip)
|
|
In reply to this post by Chris Muller-3
Please ignore. For some reason Andreas' reply didn't collect into
this conversation, so I had not seen it before I sent this. On Thu, Jan 6, 2011 at 4:46 PM, Chris Muller <[hidden email]> wrote: > Hi, yes, I put that in there because there was a test added by Matthew > Fulmer that failed if it wasn't set. I think I removed the test and > put it into ReleaseBuilderTrunk instead. > > It's perfectly fine with me to turn that change back. Ok with you, Matthew? > > - Chris > > > On Thu, Jan 6, 2011 at 2:46 PM, Bernhard Pieber <[hidden email]> wrote: >> Hi Chris, >> >> I saw that you set the #swapMouseButtons preference. Now the right mouse button is fixed on the Mac. However, I just tested the image under Windows. I am afraid now it is broken there, i.e. the right mouse shows the halos instead of the context menu. >> >> There have been numerous threads about this problem. IMHO the right solution would be: >> - Make the release VMs consistent so that one can move an image between the platforms and the right mouse button behavior stays the same without need to change preferences (or run platform specific code like Pharo does it IIRC). >> - As Mac and Linux seem to behave the same the easiest way seem to me to change the Squeak*.ini to include 3ButtonMouse=1. >> >> I fully agree with Igor on that topic: >> http://lists.squeakfoundation.org/pipermail/squeak-dev/2010-December/155896.html >> >> Cheers, >> Bernhard >> >> Am 06.01.2011 um 18:33 schrieb Chris Muller: >> >>> Hi all, I heard back from Andreas, he is fine with pushing the >>> FloatMathPlugin enhancements to 4.3. >>> >>> I loaded my latest version of KernelTests from the inbox which removes >>> FloatConsistencyTests, and have just uploaded a first 4.2 beta to >>> >>> ftp://ftp.squeak.org/4.2/ >>> >>> We can try to go-back in the trunk stream so that a continuous update >>> could work from a 10779 image, but I'm really not sure what needs to >>> be done to accomplish that or whether that's important. We have this >>> beta image from which we can at least update forward. >>> >>> We do still have some issues in the test-suite that we need help with! >>> >>> Please also test your applications with this image over the next weeks >>> to ensure they will be compatible with this new release, and also >>> flush out any final problems. >>> >>> We're getting there, thanks to all for your help. >>> >>> - Chris >>> >> >> >> > |
|
|
In reply to this post by Casey Ransberger-2
On 1/6/2011 11:54 PM, Casey Ransberger wrote:
> I like a welcome workspace. More than one is annoying to me though. That's just me though. I can see the "clean desktop" argument, and would probably prefer it, if I didn't feel that we were experiencing community growth. I think the welcome workspace is friendly to new folks. +1. Cheers, - Andreas > On Jan 6, 2011, at 2:28 PM, Chris Muller<[hidden email]> wrote: > >> Good catch. Am I reading an implicit suggestion in your note to close >> the window too, so that Squeak 4.2 launches with an empty desktop? I >> could appreciate that "clean" look. What is your preference about >> that? >> >> >> On Thu, Jan 6, 2011 at 2:30 PM, Andreas Raab<[hidden email]> wrote: >>> One comment: The "Welcome to Squeak 4.2" workspace is still all 4.1 content. >>> I think it would be best to simply copy the "What's New" section into that >>> workspace. >>> >>> Cheers, >>> - Andreas >>> >>> On 1/6/2011 6:33 PM, Chris Muller wrote: >>>> >>>> Hi all, I heard back from Andreas, he is fine with pushing the >>>> FloatMathPlugin enhancements to 4.3. >>>> >>>> I loaded my latest version of KernelTests from the inbox which removes >>>> FloatConsistencyTests, and have just uploaded a first 4.2 beta to >>>> >>>> ftp://ftp.squeak.org/4.2/ >>>> >>>> We can try to go-back in the trunk stream so that a continuous update >>>> could work from a 10779 image, but I'm really not sure what needs to >>>> be done to accomplish that or whether that's important. We have this >>>> beta image from which we can at least update forward. >>>> >>>> We do still have some issues in the test-suite that we need help with! >>>> >>>> Please also test your applications with this image over the next weeks >>>> to ensure they will be compatible with this new release, and also >>>> flush out any final problems. >>>> >>>> We're getting there, thanks to all for your help. >>>> >>>> - Chris >>>> >>>> >>> >>> >>> >> > > |
|
|
In reply to this post by Chris Muller-4
On 1/6/11 3:33 PM, "Chris Muller" <[hidden email]> wrote: > We can try to go-back in the trunk stream so that a continuous update > could work from a 10779 image, On the build of FunSqueakCog could trace back to 10740 and no problems on updating. In fact, just trying today gives me 10860. I'm on Mac and use Squeak 5.8b4 VM. Thanks and keep the good work ! Edgar |
|
|
In reply to this post by bpi
I found the reason for this. I just saved Monticello-bp.416 with a fix to the inbox.
When you do Smalltalk cleanUp: aggressive all the PackageInfos are flushed. The reason is that I implemented cleanUp: in MCPackageManager and not in its subclass MCWorkingCopy because flushObsoletePackageInfos, which I call in cleanUp:, was implemented there as well. However, if you do MCPackageManager flushObsoletePackageInfos instead of MCWorkingCopy flushObsoletePackageInfos it flushes all PackageInfos. Solution: Move flushObsoletePackageInfos and cleanUp: from MCPackageManager to MCWorkingCopy. To test: 1. Take the latest trunk image. 2. Load Monticello-bp.416 from the inbox. 3. Do Smalltalk cleanUp: true. 4. Do DependencyBrowser open. Note that all the packages are still there. 5. Run PackageDependencyTest. Only testMorphic and testSUnitGUI remain. (A fix for testSUnitGUI is in the inbox. ;-) Cheers, Bernhard Am 06.01.2011 um 20:56 schrieb Bernhard Pieber: > Am 06.01.2011 um 18:33 schrieb Chris Muller: >> Please also test your applications with this image over the next weeks >> to ensure they will be compatible with this new release, and also >> flush out any final problems. > I think there is something wrong with the PackageDependencyTest in the beta image. I did not understand why testSunitUI does not fail without my fixes and looked into it. All 35 tests pass but it runs very fast. In the trunk image it takes several seconds. When I do DependencyBrowser open it is completely empty. > > Cheers, > Bernhard |
|
|
Thanks! It's all merged.
BTW, I was shocked to learn of that DependencyBrowser! Man oh man that is great! On Fri, Jan 7, 2011 at 4:04 PM, Bernhard Pieber <[hidden email]> wrote: > I found the reason for this. I just saved Monticello-bp.416 with a fix to the inbox. > > When you do Smalltalk cleanUp: aggressive all the PackageInfos are flushed. The reason is that I implemented cleanUp: in MCPackageManager and not in its subclass MCWorkingCopy because flushObsoletePackageInfos, which I call in cleanUp:, was implemented there as well. However, if you do MCPackageManager flushObsoletePackageInfos instead of MCWorkingCopy flushObsoletePackageInfos it flushes all PackageInfos. > > Solution: Move flushObsoletePackageInfos and cleanUp: from MCPackageManager to MCWorkingCopy. > > To test: > 1. Take the latest trunk image. > 2. Load Monticello-bp.416 from the inbox. > 3. Do Smalltalk cleanUp: true. > 4. Do DependencyBrowser open. Note that all the packages are still there. > 5. Run PackageDependencyTest. Only testMorphic and testSUnitGUI remain. (A fix for testSUnitGUI is in the inbox. ;-) > > Cheers, > Bernhard > > Am 06.01.2011 um 20:56 schrieb Bernhard Pieber: > >> Am 06.01.2011 um 18:33 schrieb Chris Muller: >>> Please also test your applications with this image over the next weeks >>> to ensure they will be compatible with this new release, and also >>> flush out any final problems. >> I think there is something wrong with the PackageDependencyTest in the beta image. I did not understand why testSunitUI does not fail without my fixes and looked into it. All 35 tests pass but it runs very fast. In the trunk image it takes several seconds. When I do DependencyBrowser open it is completely empty. >> >> Cheers, >> Bernhard > > > |
|
|
Oh! That looks to be an extremely useful tool!
One thing that it can't do is pick up _missing_ dependencies: this fileout of Islands that I'm playing uses some classes that are missing from the image. What would be super great is to have an "Undeclared" item in the second pane, that would hold all those dependencies that are missing from the image. (I can imagine dependencies difficult of even impossible to pick up - (Smalltalk at: ('My', 'Foo') asSymbol) new - but I'm looking for useful things, not perfect things.) frank On 2011/01/08 22:12, Chris Muller wrote: > Thanks! It's all merged. > > BTW, I was shocked to learn of that DependencyBrowser! Man oh man > that is great! > > > On Fri, Jan 7, 2011 at 4:04 PM, Bernhard Pieber<[hidden email]> wrote: >> I found the reason for this. I just saved Monticello-bp.416 with a fix to the inbox. >> >> When you do Smalltalk cleanUp: aggressive all the PackageInfos are flushed. The reason is that I implemented cleanUp: in MCPackageManager and not in its subclass MCWorkingCopy because flushObsoletePackageInfos, which I call in cleanUp:, was implemented there as well. However, if you do MCPackageManager flushObsoletePackageInfos instead of MCWorkingCopy flushObsoletePackageInfos it flushes all PackageInfos. >> >> Solution: Move flushObsoletePackageInfos and cleanUp: from MCPackageManager to MCWorkingCopy. >> >> To test: >> 1. Take the latest trunk image. >> 2. Load Monticello-bp.416 from the inbox. >> 3. Do Smalltalk cleanUp: true. >> 4. Do DependencyBrowser open. Note that all the packages are still there. >> 5. Run PackageDependencyTest. Only testMorphic and testSUnitGUI remain. (A fix for testSUnitGUI is in the inbox. ;-) >> >> Cheers, >> Bernhard >> >> Am 06.01.2011 um 20:56 schrieb Bernhard Pieber: >> >>> Am 06.01.2011 um 18:33 schrieb Chris Muller: >>>> Please also test your applications with this image over the next weeks >>>> to ensure they will be compatible with this new release, and also >>>> flush out any final problems. >>> I think there is something wrong with the PackageDependencyTest in the beta image. I did not understand why testSunitUI does not fail without my fixes and looked into it. All 35 tests pass but it runs very fast. In the trunk image it takes several seconds. When I do DependencyBrowser open it is completely empty. >>> >>> Cheers, >>> Bernhard >> >> >> > > > |
«
Return to Squeak - Dev
|
1 view|%1 views
| Free forum by Nabble | Edit this page |