[Glass] Updating Zinc for Gemstone

Dale,

While I am getting my hands back onto Zinc, I noticed you pulled in Zinc 2.4.3 in the master branch on glassdb/zinc but never merged this with the gemstone branches, right?
Just asking the question, since I am then busy doing just that.

btw, we might need a way to tag/comment where gemstone-specific changes were made. Syncing latest changes from Pharo is a pain…
Method conflicts are showing up where no conflict exists and I am puzzled if I am loosing changes of the port or if it’s just noise.

Johan
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

Johan,

I thought everything was merged in a in good shape .... but let me double check...

Dale

----- Original Message -----
| From: "Johan Brichau" <[hidden email]>
| To: [hidden email]
| Sent: Sunday, November 17, 2013 8:42:57 AM
| Subject: [Glass] Updating Zinc for Gemstone
|
| Dale,
|
| While I am getting my hands back onto Zinc, I noticed you pulled in
| Zinc 2.4.3 in the master branch on glassdb/zinc but never merged
| this with the gemstone branches, right?
| Just asking the question, since I am then busy doing just that.
|
| btw, we might need a way to tag/comment where gemstone-specific
| changes were made. Syncing latest changes from Pharo is a pain…
| Method conflicts are showing up where no conflict exists and I am
| puzzled if I am loosing changes of the port or if it’s just noise.
|
| Johan
| _______________________________________________
| Glass mailing list
| [hidden email]
| http://lists.gemtalksystems.com/mailman/listinfo/glass
|
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

Oh wait ...that was my ston work ... too many different projects ..

Yes, it does look like the 2.4.3 stuff was not merged into gemstone2.4 branch nor the gemstone3.1 branch ...

I don't think we need to tag the gemstone-specific changes, because git handles that for us ... by pulling svens code onto our master branch and then doing platform specific branches git will manage the methods that we've changed for the port ... if we want to publish bugfixes back to sven, then we cherry pick into a topic branch off of the master branch and then do a pull request against sven's repo ... when he merges the pull request, we can pull from his master to pick up the changes, then we merge from our master into the gemstone2.4 and gemstone3.1 branches ...

Could you point to me to some examples of "Method conflicts are showing up where no conflict exists" I just went through a major reshuffling of code with ston, so I might have seen something like this already ...

Dale

----- Original Message -----
| From: "Dale K. Henrichs" <[hidden email]>
| To: "Johan Brichau" <[hidden email]>
| Cc: [hidden email]
| Sent: Sunday, November 17, 2013 10:20:30 AM
| Subject: Re: [Glass] Updating Zinc for Gemstone
|
| Johan,
|
| I thought everything was merged in a in good shape .... but let me
| double check...
|
| Dale
|
| ----- Original Message -----
| | From: "Johan Brichau" <[hidden email]>
| | To: [hidden email]
| | Sent: Sunday, November 17, 2013 8:42:57 AM
| | Subject: [Glass] Updating Zinc for Gemstone
| |
| | Dale,
| |
| | While I am getting my hands back onto Zinc, I noticed you pulled in
| | Zinc 2.4.3 in the master branch on glassdb/zinc but never merged
| | this with the gemstone branches, right?
| | Just asking the question, since I am then busy doing just that.
| |
| | btw, we might need a way to tag/comment where gemstone-specific
| | changes were made. Syncing latest changes from Pharo is a pain…
| | Method conflicts are showing up where no conflict exists and I am
| | puzzled if I am loosing changes of the port or if it’s just noise.
| |
| | Johan
| | _______________________________________________
| | Glass mailing list
| | [hidden email]
| | http://lists.gemtalksystems.com/mailman/listinfo/glass
| |
|
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

Note that in the glass_exp branch I was moving the extension methods needed into the GLASS1 project[1]

Dale

[1] https://github.com/glassdb/glass

----- Original Message -----
| From: "Dale K. Henrichs" <[hidden email]>
| To: "Johan Brichau" <[hidden email]>
| Cc: [hidden email]
| Sent: Sunday, November 17, 2013 10:35:00 AM
| Subject: Re: [Glass] Updating Zinc for Gemstone
|
| Oh wait ...that was my ston work ... too many different projects ..
|
| Yes, it does look like the 2.4.3 stuff was not merged into
| gemstone2.4 branch nor the gemstone3.1 branch ...
|
| I don't think we need to tag the gemstone-specific changes, because
| git handles that for us ... by pulling svens code onto our master
| branch and then doing platform specific branches git will manage the
| methods that we've changed for the port ... if we want to publish
| bugfixes back to sven, then we cherry pick into a topic branch off
| of the master branch and then do a pull request against sven's repo
| ... when he merges the pull request, we can pull from his master to
| pick up the changes, then we merge from our master into the
| gemstone2.4 and gemstone3.1 branches ...
|
| Could you point to me to some examples of "Method conflicts are
| showing up where no conflict exists" I just went through a major
| reshuffling of code with ston, so I might have seen something like
| this already ...
|
| Dale
|
| ----- Original Message -----
| | From: "Dale K. Henrichs" <[hidden email]>
| | To: "Johan Brichau" <[hidden email]>
| | Cc: [hidden email]
| | Sent: Sunday, November 17, 2013 10:20:30 AM
| | Subject: Re: [Glass] Updating Zinc for Gemstone
| |
| | Johan,
| |
| | I thought everything was merged in a in good shape .... but let me
| | double check...
| |
| | Dale
| |
| | ----- Original Message -----
| | | From: "Johan Brichau" <[hidden email]>
| | | To: [hidden email]
| | | Sent: Sunday, November 17, 2013 8:42:57 AM
| | | Subject: [Glass] Updating Zinc for Gemstone
| | |
| | | Dale,
| | |
| | | While I am getting my hands back onto Zinc, I noticed you pulled
| | | in
| | | Zinc 2.4.3 in the master branch on glassdb/zinc but never merged
| | | this with the gemstone branches, right?
| | | Just asking the question, since I am then busy doing just that.
| | |
| | | btw, we might need a way to tag/comment where gemstone-specific
| | | changes were made. Syncing latest changes from Pharo is a pain…
| | | Method conflicts are showing up where no conflict exists and I am
| | | puzzled if I am loosing changes of the port or if it’s just
| | | noise.
| | |
| | | Johan
| | | _______________________________________________
| | | Glass mailing list
| | | [hidden email]
| | | http://lists.gemtalksystems.com/mailman/listinfo/glass
| | |
| |
|
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

On 17 Nov 2013, at 19:35, Dale K. Henrichs <[hidden email]> wrote:

> Could you point to me to some examples of "Method conflicts are showing up where no conflict exists" I just went through a major reshuffling of code with ston, so I might have seen something like this already ...

I'm currently busy merging and fixing in my fork. I cannot remember exactly which methods I was confused about yesterday, but today I hit the following problem:

The class ZnChunkedReadStreamTests was removed entirely in the master branch. Because an #expectedFailures method was added to that class in the gs branch, the merge showed a conflict on the methodProperties.json. While running through the changes and conflicts, it was not apparent to me that the class was actually removed from the master branch. So I made a wrong merge, leading to a corrupt filetree format.

Now, I get built errors on travis (https://travis-ci.org/jbrichau/zinc) I cannot make sense of, since the load happens just fine locally. On travis I get:

TravisCIHarness
  value: #( 'BaselineOfZinc' )
  value: 'TravisCISuccess.txt'
  value: 'TravisCIFailure.txt'.
%
--transcript--'travis---->travisCI.st'
--transcript--'...RETRY->BaselineOfZinc'
--transcript--'...RETRY->BaselineOfZinc'
--transcript--'gofer repository error: #''GoferRepositoryError''...ignoring'
--transcript--'...FAILED->BaselineOfZinc'
-----------------------------------------------------
GemStone: Error         Nonfatal
#'MetacelloPackageSpecResolutionError'
Error Category: [User] Number: 1 Arg Count: 1
Arg 1: MetacelloPackageSpecResolutionError
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

I'm taking a look now ...

----- Original Message -----
| From: "Johan Brichau" <[hidden email]>
| To: "Dale K. Henrichs" <[hidden email]>
| Cc: [hidden email]
| Sent: Monday, November 18, 2013 11:32:24 AM
| Subject: Re: [Glass] Updating Zinc for Gemstone
|
|
| On 17 Nov 2013, at 19:35, Dale K. Henrichs
| <[hidden email]> wrote:
|
| > Could you point to me to some examples of "Method conflicts are
| > showing up where no conflict exists" I just went through a major
| > reshuffling of code with ston, so I might have seen something like
| > this already ...
|
| I'm currently busy merging and fixing in my fork. I cannot remember
| exactly which methods I was confused about yesterday, but today I
| hit the following problem:
|
| The class ZnChunkedReadStreamTests was removed entirely in the master
| branch. Because an #expectedFailures method was added to that class
| in the gs branch, the merge showed a conflict on the
| methodProperties.json. While running through the changes and
| conflicts, it was not apparent to me that the class was actually
| removed from the master branch. So I made a wrong merge, leading to
| a corrupt filetree format.
|
| Now, I get built errors on travis
| (https://travis-ci.org/jbrichau/zinc) I cannot make sense of, since
| the load happens just fine locally. On travis I get:
|
| TravisCIHarness
|   value: #( 'BaselineOfZinc' )
|   value: 'TravisCISuccess.txt'
|   value: 'TravisCIFailure.txt'.
| %
| --transcript--'travis---->travisCI.st'
| --transcript--'...RETRY->BaselineOfZinc'
| --transcript--'...RETRY->BaselineOfZinc'
| --transcript--'gofer repository error:
| #''GoferRepositoryError''...ignoring'
| --transcript--'...FAILED->BaselineOfZinc'
| -----------------------------------------------------
| GemStone: Error         Nonfatal
| #'MetacelloPackageSpecResolutionError'
| Error Category: [User] Number: 1 Arg Count: 1
| Arg 1: MetacelloPackageSpecResolutionError
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

I've merged your gemstone2.4 into my zinc repo (travis_trouble branch) and I'll see if I can get to the bottom of the baseline load failure ... it's a "familiar problem" but I'm failing to remember the "familiar solution":)

Dale

----- Original Message -----
| From: "Johan Brichau" <[hidden email]>
| To: "Dale K. Henrichs" <[hidden email]>
| Cc: [hidden email]
| Sent: Monday, November 18, 2013 11:32:24 AM
| Subject: Re: [Glass] Updating Zinc for Gemstone
|
|
| On 17 Nov 2013, at 19:35, Dale K. Henrichs
| <[hidden email]> wrote:
|
| > Could you point to me to some examples of "Method conflicts are
| > showing up where no conflict exists" I just went through a major
| > reshuffling of code with ston, so I might have seen something like
| > this already ...
|
| I'm currently busy merging and fixing in my fork. I cannot remember
| exactly which methods I was confused about yesterday, but today I
| hit the following problem:
|
| The class ZnChunkedReadStreamTests was removed entirely in the master
| branch. Because an #expectedFailures method was added to that class
| in the gs branch, the merge showed a conflict on the
| methodProperties.json. While running through the changes and
| conflicts, it was not apparent to me that the class was actually
| removed from the master branch. So I made a wrong merge, leading to
| a corrupt filetree format.
|
| Now, I get built errors on travis
| (https://travis-ci.org/jbrichau/zinc) I cannot make sense of, since
| the load happens just fine locally. On travis I get:
|
| TravisCIHarness
|   value: #( 'BaselineOfZinc' )
|   value: 'TravisCISuccess.txt'
|   value: 'TravisCIFailure.txt'.
| %
| --transcript--'travis---->travisCI.st'
| --transcript--'...RETRY->BaselineOfZinc'
| --transcript--'...RETRY->BaselineOfZinc'
| --transcript--'gofer repository error:
| #''GoferRepositoryError''...ignoring'
| --transcript--'...FAILED->BaselineOfZinc'
| -----------------------------------------------------
| GemStone: Error         Nonfatal
| #'MetacelloPackageSpecResolutionError'
| Error Category: [User] Number: 1 Arg Count: 1
| Arg 1: MetacelloPackageSpecResolutionError
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

... I turned on 3.1.0.4 just for grins and it turns out that the default Metacello loaded into 2.1.0.4 has a better error message[1]:

  --transcript--'travis---->travisCI.st'
  --transcript--'...RETRY->BaselineOfZinc'
  --transcript--'...RETRY->BaselineOfZinc'
  --transcript--'gofer repository error: ''GoferRepositoryError: string contents are ill-formed Utf8 at byte offset 76507''...ignoring'
  --transcript--'...FAILED->BaselineOfZinc'

So there's some invalid utf8 encountered while reading in the BaselineOfZinc ... will investigate further and issue a pull request when I have a solution..

Dale

[1] https://travis-ci.org/dalehenrich/zinc/jobs/14160177#L417

----- Original Message -----
| From: "Dale K. Henrichs" <[hidden email]>
| To: "Johan Brichau" <[hidden email]>
| Cc: [hidden email]
| Sent: Monday, November 18, 2013 12:33:17 PM
| Subject: Re: [Glass] Updating Zinc for Gemstone
|
| I've merged your gemstone2.4 into my zinc repo (travis_trouble
| branch) and I'll see if I can get to the bottom of the baseline load
| failure ... it's a "familiar problem" but I'm failing to remember
| the "familiar solution":)
|
| Dale
|
| ----- Original Message -----
| | From: "Johan Brichau" <[hidden email]>
| | To: "Dale K. Henrichs" <[hidden email]>
| | Cc: [hidden email]
| | Sent: Monday, November 18, 2013 11:32:24 AM
| | Subject: Re: [Glass] Updating Zinc for Gemstone
| |
| |
| | On 17 Nov 2013, at 19:35, Dale K. Henrichs
| | <[hidden email]> wrote:
| |
| | > Could you point to me to some examples of "Method conflicts are
| | > showing up where no conflict exists" I just went through a major
| | > reshuffling of code with ston, so I might have seen something
| | > like
| | > this already ...
| |
| | I'm currently busy merging and fixing in my fork. I cannot remember
| | exactly which methods I was confused about yesterday, but today I
| | hit the following problem:
| |
| | The class ZnChunkedReadStreamTests was removed entirely in the
| | master
| | branch. Because an #expectedFailures method was added to that class
| | in the gs branch, the merge showed a conflict on the
| | methodProperties.json. While running through the changes and
| | conflicts, it was not apparent to me that the class was actually
| | removed from the master branch. So I made a wrong merge, leading to
| | a corrupt filetree format.
| |
| | Now, I get built errors on travis
| | (https://travis-ci.org/jbrichau/zinc) I cannot make sense of, since
| | the load happens just fine locally. On travis I get:
| |
| | TravisCIHarness
| |   value: #( 'BaselineOfZinc' )
| |   value: 'TravisCISuccess.txt'
| |   value: 'TravisCIFailure.txt'.
| | %
| | --transcript--'travis---->travisCI.st'
| | --transcript--'...RETRY->BaselineOfZinc'
| | --transcript--'...RETRY->BaselineOfZinc'
| | --transcript--'gofer repository error:
| | #''GoferRepositoryError''...ignoring'
| | --transcript--'...FAILED->BaselineOfZinc'
| | -----------------------------------------------------
| | GemStone: Error         Nonfatal
| | #'MetacelloPackageSpecResolutionError'
| | Error Category: [User] Number: 1 Arg Count: 1
| | Arg 1: MetacelloPackageSpecResolutionError
|
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

The better error message is probably because Unicode support has been added to the base in 3.1 and the Unicode primitive is giving a good error and the Metacello code is displaying the error...

Dale

----- Original Message -----
| From: "Dale K. Henrichs" <[hidden email]>
| To: "Johan Brichau" <[hidden email]>
| Cc: [hidden email]
| Sent: Monday, November 18, 2013 12:37:33 PM
| Subject: Re: [Glass] Updating Zinc for Gemstone
|
| ... I turned on 3.1.0.4 just for grins and it turns out that the
| default Metacello loaded into 2.1.0.4 has a better error message[1]:
|
|   --transcript--'travis---->travisCI.st'
|   --transcript--'...RETRY->BaselineOfZinc'
|   --transcript--'...RETRY->BaselineOfZinc'
|   --transcript--'gofer repository error: ''GoferRepositoryError:
|   string contents are ill-formed Utf8 at byte offset
|   76507''...ignoring'
|   --transcript--'...FAILED->BaselineOfZinc'
|
| So there's some invalid utf8 encountered while reading in the
| BaselineOfZinc ... will investigate further and issue a pull request
| when I have a solution..
|
| Dale
|
| [1] https://travis-ci.org/dalehenrich/zinc/jobs/14160177#L417
|
| ----- Original Message -----
| | From: "Dale K. Henrichs" <[hidden email]>
| | To: "Johan Brichau" <[hidden email]>
| | Cc: [hidden email]
| | Sent: Monday, November 18, 2013 12:33:17 PM
| | Subject: Re: [Glass] Updating Zinc for Gemstone
| |
| | I've merged your gemstone2.4 into my zinc repo (travis_trouble
| | branch) and I'll see if I can get to the bottom of the baseline
| | load
| | failure ... it's a "familiar problem" but I'm failing to remember
| | the "familiar solution":)
| |
| | Dale
| |
| | ----- Original Message -----
| | | From: "Johan Brichau" <[hidden email]>
| | | To: "Dale K. Henrichs" <[hidden email]>
| | | Cc: [hidden email]
| | | Sent: Monday, November 18, 2013 11:32:24 AM
| | | Subject: Re: [Glass] Updating Zinc for Gemstone
| | |
| | |
| | | On 17 Nov 2013, at 19:35, Dale K. Henrichs
| | | <[hidden email]> wrote:
| | |
| | | > Could you point to me to some examples of "Method conflicts are
| | | > showing up where no conflict exists" I just went through a
| | | > major
| | | > reshuffling of code with ston, so I might have seen something
| | | > like
| | | > this already ...
| | |
| | | I'm currently busy merging and fixing in my fork. I cannot
| | | remember
| | | exactly which methods I was confused about yesterday, but today I
| | | hit the following problem:
| | |
| | | The class ZnChunkedReadStreamTests was removed entirely in the
| | | master
| | | branch. Because an #expectedFailures method was added to that
| | | class
| | | in the gs branch, the merge showed a conflict on the
| | | methodProperties.json. While running through the changes and
| | | conflicts, it was not apparent to me that the class was actually
| | | removed from the master branch. So I made a wrong merge, leading
| | | to
| | | a corrupt filetree format.
| | |
| | | Now, I get built errors on travis
| | | (https://travis-ci.org/jbrichau/zinc) I cannot make sense of,
| | | since
| | | the load happens just fine locally. On travis I get:
| | |
| | | TravisCIHarness
| | |   value: #( 'BaselineOfZinc' )
| | |   value: 'TravisCISuccess.txt'
| | |   value: 'TravisCIFailure.txt'.
| | | %
| | | --transcript--'travis---->travisCI.st'
| | | --transcript--'...RETRY->BaselineOfZinc'
| | | --transcript--'...RETRY->BaselineOfZinc'
| | | --transcript--'gofer repository error:
| | | #''GoferRepositoryError''...ignoring'
| | | --transcript--'...FAILED->BaselineOfZinc'
| | | -----------------------------------------------------
| | | GemStone: Error         Nonfatal
| | | #'MetacelloPackageSpecResolutionError'
| | | Error Category: [User] Number: 1 Arg Count: 1
| | | Arg 1: MetacelloPackageSpecResolutionError
| |
|
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

Johan,

The pull request[1] gets past the loading problem ... the ill-formed utf8 character wa since in the version file and there was an older fileTree bug[2] where gemstone was not encoding the monticello meta data files in utf8.

Dale

[1] https://github.com/jbrichau/zinc/pull/4
[2] https://github.com/dalehenrich/filetree/issues/72

----- Original Message -----
| From: "Dale K. Henrichs" <[hidden email]>
| To: "Johan Brichau" <[hidden email]>
| Cc: [hidden email]
| Sent: Monday, November 18, 2013 12:37:33 PM
| Subject: Re: [Glass] Updating Zinc for Gemstone
|
| ... I turned on 3.1.0.4 just for grins and it turns out that the
| default Metacello loaded into 2.1.0.4 has a better error message[1]:
|
|   --transcript--'travis---->travisCI.st'
|   --transcript--'...RETRY->BaselineOfZinc'
|   --transcript--'...RETRY->BaselineOfZinc'
|   --transcript--'gofer repository error: ''GoferRepositoryError:
|   string contents are ill-formed Utf8 at byte offset
|   76507''...ignoring'
|   --transcript--'...FAILED->BaselineOfZinc'
|
| So there's some invalid utf8 encountered while reading in the
| BaselineOfZinc ... will investigate further and issue a pull request
| when I have a solution..
|
| Dale
|
| [1] https://travis-ci.org/dalehenrich/zinc/jobs/14160177#L417
|
| ----- Original Message -----
| | From: "Dale K. Henrichs" <[hidden email]>
| | To: "Johan Brichau" <[hidden email]>
| | Cc: [hidden email]
| | Sent: Monday, November 18, 2013 12:33:17 PM
| | Subject: Re: [Glass] Updating Zinc for Gemstone
| |
| | I've merged your gemstone2.4 into my zinc repo (travis_trouble
| | branch) and I'll see if I can get to the bottom of the baseline
| | load
| | failure ... it's a "familiar problem" but I'm failing to remember
| | the "familiar solution":)
| |
| | Dale
| |
| | ----- Original Message -----
| | | From: "Johan Brichau" <[hidden email]>
| | | To: "Dale K. Henrichs" <[hidden email]>
| | | Cc: [hidden email]
| | | Sent: Monday, November 18, 2013 11:32:24 AM
| | | Subject: Re: [Glass] Updating Zinc for Gemstone
| | |
| | |
| | | On 17 Nov 2013, at 19:35, Dale K. Henrichs
| | | <[hidden email]> wrote:
| | |
| | | > Could you point to me to some examples of "Method conflicts are
| | | > showing up where no conflict exists" I just went through a
| | | > major
| | | > reshuffling of code with ston, so I might have seen something
| | | > like
| | | > this already ...
| | |
| | | I'm currently busy merging and fixing in my fork. I cannot
| | | remember
| | | exactly which methods I was confused about yesterday, but today I
| | | hit the following problem:
| | |
| | | The class ZnChunkedReadStreamTests was removed entirely in the
| | | master
| | | branch. Because an #expectedFailures method was added to that
| | | class
| | | in the gs branch, the merge showed a conflict on the
| | | methodProperties.json. While running through the changes and
| | | conflicts, it was not apparent to me that the class was actually
| | | removed from the master branch. So I made a wrong merge, leading
| | | to
| | | a corrupt filetree format.
| | |
| | | Now, I get built errors on travis
| | | (https://travis-ci.org/jbrichau/zinc) I cannot make sense of,
| | | since
| | | the load happens just fine locally. On travis I get:
| | |
| | | TravisCIHarness
| | |   value: #( 'BaselineOfZinc' )
| | |   value: 'TravisCISuccess.txt'
| | |   value: 'TravisCIFailure.txt'.
| | | %
| | | --transcript--'travis---->travisCI.st'
| | | --transcript--'...RETRY->BaselineOfZinc'
| | | --transcript--'...RETRY->BaselineOfZinc'
| | | --transcript--'gofer repository error:
| | | #''GoferRepositoryError''...ignoring'
| | | --transcript--'...FAILED->BaselineOfZinc'
| | | -----------------------------------------------------
| | | GemStone: Error         Nonfatal
| | | #'MetacelloPackageSpecResolutionError'
| | | Error Category: [User] Number: 1 Arg Count: 1
| | | Arg 1: MetacelloPackageSpecResolutionError
| |
|
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

Hi Dale,

I have the same problem with ill-formed utf8 in the Seaside31 build.

I must say I don’t understand this: the load happens correctly with my local repository but not on Travis?

Besides that, how does one find these ill-formed utf8 files?

Johan

On 18 Nov 2013, at 22:21, Dale K. Henrichs <[hidden email]> wrote:

_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

----- Original Message -----
| From: "Johan Brichau" <[hidden email]>
| To: "Dale K. Henrichs" <[hidden email]>
| Cc: [hidden email]
| Sent: Saturday, November 30, 2013 5:40:40 AM
| Subject: Re: [Glass] Updating Zinc for Gemstone
|
| Hi Dale,
|
| I have the same problem with ill-formed utf8 in the Seaside31 build.
|
| I must say I don’t understand this: the load happens correctly with
| my local repository but not on Travis?

I'm pretty sure that travis uses the latest version of FileTree. Do you happen to be using an older version of FileTree with your local respository?

|
| Besides that, how does one find these ill-formed utf8 files?

I found the problem when I attempted to look at the repository with my tODE instance (walkback while scanning the repository I think) ... I didn't try looking with GemTools, but if you are using an older version of FileTree, you might not see the problem just junk characters in a comment for an older version...

I'll see if I can find the problem package(s) for Seaside31 and I'll pay closer attention to what I do and also try some things with GemTools ...

Dale
|
| Johan
|
| On 18 Nov 2013, at 22:21, Dale K. Henrichs
| <[hidden email]> wrote:
|
| > Johan,
| >
| > The pull request[1] gets past the loading problem ... the
| > ill-formed utf8 character wa since in the version file and there
| > was an older fileTree bug[2] where gemstone was not encoding the
| > monticello meta data files in utf8.
| >
| > Dale
| >
| > [1] https://github.com/jbrichau/zinc/pull/4
| > [2] https://github.com/dalehenrich/filetree/issues/72
| >
| > ----- Original Message -----
| > | From: "Dale K. Henrichs" <[hidden email]>
| > | To: "Johan Brichau" <[hidden email]>
| > | Cc: [hidden email]
| > | Sent: Monday, November 18, 2013 12:37:33 PM
| > | Subject: Re: [Glass] Updating Zinc for Gemstone
| > |
| > | ... I turned on 3.1.0.4 just for grins and it turns out that the
| > | default Metacello loaded into 2.1.0.4 has a better error
| > | message[1]:
| > |
| > |   --transcript--'travis---->travisCI.st'
| > |   --transcript--'...RETRY->BaselineOfZinc'
| > |   --transcript--'...RETRY->BaselineOfZinc'
| > |   --transcript--'gofer repository error: ''GoferRepositoryError:
| > |   string contents are ill-formed Utf8 at byte offset
| > |   76507''...ignoring'
| > |   --transcript--'...FAILED->BaselineOfZinc'
| > |
| > | So there's some invalid utf8 encountered while reading in the
| > | BaselineOfZinc ... will investigate further and issue a pull
| > | request
| > | when I have a solution..
| > |
| > | Dale
| > |
| > | [1] https://travis-ci.org/dalehenrich/zinc/jobs/14160177#L417
| > |
| > | ----- Original Message -----
| > | | From: "Dale K. Henrichs" <[hidden email]>
| > | | To: "Johan Brichau" <[hidden email]>
| > | | Cc: [hidden email]
| > | | Sent: Monday, November 18, 2013 12:33:17 PM
| > | | Subject: Re: [Glass] Updating Zinc for Gemstone
| > | |
| > | | I've merged your gemstone2.4 into my zinc repo (travis_trouble
| > | | branch) and I'll see if I can get to the bottom of the baseline
| > | | load
| > | | failure ... it's a "familiar problem" but I'm failing to
| > | | remember
| > | | the "familiar solution":)
| > | |
| > | | Dale
| > | |
| > | | ----- Original Message -----
| > | | | From: "Johan Brichau" <[hidden email]>
| > | | | To: "Dale K. Henrichs" <[hidden email]>
| > | | | Cc: [hidden email]
| > | | | Sent: Monday, November 18, 2013 11:32:24 AM
| > | | | Subject: Re: [Glass] Updating Zinc for Gemstone
| > | | |
| > | | |
| > | | | On 17 Nov 2013, at 19:35, Dale K. Henrichs
| > | | | <[hidden email]> wrote:
| > | | |
| > | | | > Could you point to me to some examples of "Method conflicts
| > | | | > are
| > | | | > showing up where no conflict exists" I just went through a
| > | | | > major
| > | | | > reshuffling of code with ston, so I might have seen
| > | | | > something
| > | | | > like
| > | | | > this already ...
| > | | |
| > | | | I'm currently busy merging and fixing in my fork. I cannot
| > | | | remember
| > | | | exactly which methods I was confused about yesterday, but
| > | | | today I
| > | | | hit the following problem:
| > | | |
| > | | | The class ZnChunkedReadStreamTests was removed entirely in
| > | | | the
| > | | | master
| > | | | branch. Because an #expectedFailures method was added to that
| > | | | class
| > | | | in the gs branch, the merge showed a conflict on the
| > | | | methodProperties.json. While running through the changes and
| > | | | conflicts, it was not apparent to me that the class was
| > | | | actually
| > | | | removed from the master branch. So I made a wrong merge,
| > | | | leading
| > | | | to
| > | | | a corrupt filetree format.
| > | | |
| > | | | Now, I get built errors on travis
| > | | | (https://travis-ci.org/jbrichau/zinc) I cannot make sense of,
| > | | | since
| > | | | the load happens just fine locally. On travis I get:
| > | | |
| > | | | TravisCIHarness
| > | | |   value: #( 'BaselineOfZinc' )
| > | | |   value: 'TravisCISuccess.txt'
| > | | |   value: 'TravisCIFailure.txt'.
| > | | | %
| > | | | --transcript--'travis---->travisCI.st'
| > | | | --transcript--'...RETRY->BaselineOfZinc'
| > | | | --transcript--'...RETRY->BaselineOfZinc'
| > | | | --transcript--'gofer repository error:
| > | | | #''GoferRepositoryError''...ignoring'
| > | | | --transcript--'...FAILED->BaselineOfZinc'
| > | | | -----------------------------------------------------
| > | | | GemStone: Error         Nonfatal
| > | | | #'MetacelloPackageSpecResolutionError'
| > | | | Error Category: [User] Number: 1 Arg Count: 1
| > | | | Arg 1: MetacelloPackageSpecResolutionError
| > | |
| > |
|
|
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

Johan,

I checked out the commit that was right before the one where you fixed the issue.

It looks like JQuery-Tests-Core is the culprit (at least this was the first one to fail in tODE)... In GemTools I had to open a repository browser and click on the package name in the left pane and then click on the package version in the right pane ... and boom...

This does happen to be the same repository that I've got tODE loaded in and the error is coming from the most recent version of FileTree (github://dalehenrich/filetree:gemstone2.4/repository)...

You get a pretty good error message with the character offset into the file:

  a ArgumentError occurred (error 2718), string contents are ill-formed Utf8 at byte offset 2714, (name 'JQuery-Tests-Core-pmm.138' message '- portability' id '191175d4-1291-4715-9431-4e0a0ba69f0a' date '09/13/2013' time '10:56:50' author 'pmm' ancestors ((name 'JQuery-Tests-Core-pmm.137' message ......

As to why you haven't seen the problem in your image, according to the version file history[1] the package was brought into the Seaside31 repo on September 29th and the bug was fixed 6 months ago, so the most likely explanation is that the version of FileTree is/was old, but it isn't necessarily too far-fetched that FilTree hasn't been updated form the version that shipped with the 3.1.0.x release (which was old). Looks like a Metacello baseline load will update the version of FileTree...

Dale

[1] https://github.com/glassdb/Seaside31/commits/master/repository/JQuery-Tests-Core.package/monticello.meta/version
[2] https://github.com/dalehenrich/filetree/issues/72
_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass