[Pharo-project] About FileSystem
Hi guys Igor convinced me to publish filesystem into the squeaksource Filesystem so that others people can not feel that we are stealing it from colin. Now since we want it as a fondation for the our system, we should invest on it. and this is what I'm doing. So here is what I did - fetched all the versions from colin, lukas, pharotaskforces - pushed all the versions to squeaksource/Filesystem - pushed all the versions to pharoTaskForces Now I will publish to squeaksource/Filesystem. If people want to join this is open and free. Stef
I tried to publish a new configurationOfFilesystem but after 15 min I stopped because I could not make it load. For me there is something wrong with the interface of metacello because I can never remember it version: '1.0.2' did not load it while it defined as version: #'1.0.2' and of course I tried string and symobols.... And I got a bug in the decompiler that crashed my image probably here [tempVector remoteTemps size = maybeTVTag value size]]] assert] and so I lost everything... popIntoTemporaryVariable: offset | maybeTVTag tempVector start | maybeTVTag := stack last. ((maybeTVTag isMemberOf: Association) and: [maybeTVTag key == #pushNewArray]) ifTrue: [blockStartsToTempVars notNil "implies we were intialized with temp names." ifTrue: "Use the provided temps" [[(tempVector := tempVars at: offset + 1 ifAbsent: [ParseNode basicNew]) isTemp and: [tempVector isIndirectTempVector and: [tempVector remoteTemps size = maybeTVTag value size]]] assert] ifFalse: "Synthesize some remote temps" [tempVector := maybeTVTag value. offset + 1 <= tempVars size ifTrue: [start := 2. tempVector at: 1 put: (tempVars at: offset + 1)] ifFalse: [tempVars := (Array new: offset + 1) replaceFrom: 1 to: tempVars size with: tempVars. start := 1]. start to: tempVector size do: [:i| tempVector at: i put: (constructor codeTemp: numLocalTemps + offset + i - 1 named: 't', (tempVarCount + i) printString)]. tempVars at: offset + 1 put: (constructor codeRemoteTemp: offset + 1 remoteTemps: tempVector)]. tempVarCount := tempVarCount + maybeTVTag value size. stack removeLast. ^self]. self pushTemporaryVariable: offset; doStore: statements On Feb 3, 2011, at 10:43 AM, Stéphane Ducasse wrote:
Hi guys
Igor convinced me to publish filesystem into the squeaksource Filesystem so that others people can not feel that we are stealing it from colin. Now since we want it as a fondation for the our system, we should invest on it. and this is what I'm doing.
So here is what I did - fetched all the versions from colin, lukas, pharotaskforces - pushed all the versions to squeaksource/Filesystem - pushed all the versions to pharoTaskForces
Now I will publish to squeaksource/Filesystem.
If people want to join this is open and free.
Stef
On Feb 3, 2011, at 1:08 PM, Stéphane Ducasse wrote:
I tried to publish a new configurationOfFilesystem but after 15 min I stopped because I could not make it load. For me there is something wrong with the interface of metacello because I can never remember it version: '1.0.2' did not load it while it defined as version: #'1.0.2' and of course I tried string and symobols....
Stef, I recommend that you try to use the validator: (MetacelloToolBox validateConfiguration: ConfigurationOfFilesystem) inspect The validator should complain about using #'1.0.2' or '1.0.2' incorrectly. If the issue printString isn't enough you should be able to get more info from the #reasonCode: MetacelloToolBox descriptionForValidationReasonCode: <reasonCode from validation issue> If the validator doesn't provide a good enough explanation, I'm interested in improving it. If you run into a problem when you reconstruct the configuration and the validator doesn't give you good information, then please feel free to ship me a copy of the configuration so I can be the human validator - which in turn will spur me on to improve the validator:) I extend this invitation to anyone having trouble with Metacello. Dale
Ok I'm trying I have version102: spec <version: '1.0.2' imports: #('1.0-baseline') > spec for: #common do: [ spec blessing: #release. spec package: 'FileSystem']. (ConfigurationOfFilesystem project version: '1.0.2') load worked now (MetacelloToolBox validateConfiguration: ConfigurationOfFilesystem) inspect Critical Warning: No version specified for the package 'Filesystem' in version '1.0.1' { noVersionSpecified } [ #validateVersionSpec: ] Critical Warning: The name of package 'FileSystem' does not match the file name 'Filesystem-cwp.63' in version '1.0.1' { packageNameMismatch } [ #validateVersionSpec: ] Critical Warning: No version specified for the package 'Filesystem' in version '1.0.2' { noVersionSpecified } [ #validateVersionSpec: ] I was confused why this is a critical warning? MetacelloToolBox descriptionForValidationReasonCode: #noVersionSpecified I got a DNU at:at: Stef On Feb 3, 2011, at 10:18 PM, Dale Henrichs wrote:
On Feb 3, 2011, at 1:08 PM, Stéphane Ducasse wrote:
I tried to publish a new configurationOfFilesystem but after 15 min I stopped because I could not make it load. For me there is something wrong with the interface of metacello because I can never remember it version: '1.0.2' did not load it while it defined as version: #'1.0.2' and of course I tried string and symobols....
Stef,
I recommend that you try to use the validator:
(MetacelloToolBox validateConfiguration: ConfigurationOfFilesystem) inspect
The validator should complain about using #'1.0.2' or '1.0.2' incorrectly. If the issue printString isn't enough you should be able to get more info from the #reasonCode:
MetacelloToolBox descriptionForValidationReasonCode: <reasonCode from validation issue>
If the validator doesn't provide a good enough explanation, I'm interested in improving it.
If you run into a problem when you reconstruct the configuration and the validator doesn't give you good information, then please feel free to ship me a copy of the configuration so I can be the human validator - which in turn will spur me on to improve the validator:)
I extend this invitation to anyone having trouble with Metacello.
Dale
On Feb 4, 2011, at 12:12 AM, Stéphane Ducasse wrote:
Ok I'm trying
I have
version102: spec <version: '1.0.2' imports: #('1.0-baseline') >
spec for: #common do: [ spec blessing: #release. spec package: 'FileSystem'].
(ConfigurationOfFilesystem project version: '1.0.2') load
worked now
(MetacelloToolBox validateConfiguration: ConfigurationOfFilesystem) inspect
Critical Warning: No version specified for the package 'Filesystem' in version '1.0.1' { noVersionSpecified } [ #validateVersionSpec: ] Critical Warning: The name of package 'FileSystem' does not match the file name 'Filesystem-cwp.63' in version '1.0.1' { packageNameMismatch } [ #validateVersionSpec: ] Critical Warning: No version specified for the package 'Filesystem' in version '1.0.2' { noVersionSpecified } [ #validateVersionSpec: ]
I was confused why this is a critical warning?
MetacelloToolBox descriptionForValidationReasonCode: #noVersionSpecified
I got a DNU at:at:
Ah yes, the at:at: problem .. I have had a fix for that in 1.0-beta.28.3 for awhile but got busy and neglected to push out a new version ... too many things going on:( The description for #noVersionSpecified is: 'no version defined for the project reference or package. The version specified in the baseline or the latest version of the project or package in the repository will be used.' and the description for #packageNameMismatch is: 'the name in the packageSpec does not match the name of the mcz file.' You haven't provided the baseline spec or the spec for 1.0.1, but it looks like you must have used the name: Filesystem in the baseline, while you are using the name FileSystem (with a capital S) in the 1.0.2 spec and in the 1.0.1 spec. Perhaps you have renamed the package from Filesystem... to FileSystem...? #noVersionSpecified is a criticalWarning, because a version spec should have explicit versions specified for all packages and projects, otherwise you don't have a complete specification. As the 1.0.2 spec stands the latest mz file in the repository will be loaded. I'll try to push out 1.0-beta.28.3 this weekend:) Dale
thanks dale. I really like the idea of your validation tool.
MetacelloToolBox descriptionForValidationReasonCode: #noVersionSpecified
I got a DNU at:at:
Ah yes, the at:at: problem .. I have had a fix for that in 1.0-beta.28.3 for awhile but got busy and neglected to push out a new version ... too many things going on:(
The description for #noVersionSpecified is:
'no version defined for the project reference or package. The version specified in the baseline or the latest version of the project or package in the repository will be used.'
and the description for #packageNameMismatch is:
'the name in the packageSpec does not match the name of the mcz file.'
You haven't provided the baseline spec or the spec for 1.0.1, but it looks like you must have used the name:
Filesystem
in the baseline, while you are using the name
FileSystem
(with a capital S) in the 1.0.2 spec and in the 1.0.1 spec. Perhaps you have renamed the package from Filesystem... to FileSystem...?
#noVersionSpecified is a criticalWarning, because a version spec should have explicit versions specified for all packages and projects, otherwise you don't have a complete specification. As the 1.0.2 spec stands the latest mz file in the repository will be loaded.
I'll try to push out 1.0-beta.28.3 this weekend:)
Dale
On 02/04/2011 12:06 PM, Stéphane Ducasse wrote:
thanks dale. I really like the idea of your validation tool.
Yeah, but I think I must need a validator for my validator tool:) Dale
MetacelloToolBox descriptionForValidationReasonCode: #noVersionSpecified
I got a DNU at:at:
Ah yes, the at:at: problem .. I have had a fix for that in 1.0-beta.28.3 for awhile but got busy and neglected to push out a new version ... too many things going on:(
The description for #noVersionSpecified is:
'no version defined for the project reference or package. The version specified in the baseline or the latest version of the project or package in the repository will be used.'
and the description for #packageNameMismatch is:
'the name in the packageSpec does not match the name of the mcz file.'
You haven't provided the baseline spec or the spec for 1.0.1, but it looks like you must have used the name:
Filesystem
in the baseline, while you are using the name
FileSystem
(with a capital S) in the 1.0.2 spec and in the 1.0.1 spec. Perhaps you have renamed the package from Filesystem... to FileSystem...?
#noVersionSpecified is a criticalWarning, because a version spec should have explicit versions specified for all packages and projects, otherwise you don't have a complete specification. As the 1.0.2 spec stands the latest mz file in the repository will be loaded.
I'll try to push out 1.0-beta.28.3 this weekend:)
Dale
On Feb 4, 2011, at 9:23 32PM, Dale Henrichs wrote:
On 02/04/2011 12:06 PM, Stéphane Ducasse wrote:
thanks dale. I really like the idea of your validation tool.
Yeah, but I think I must need a validator for my validator tool:)
Dale
MetaMetaMetacello? ;) Cheers, Henry
participants (3)
-
Dale Henrichs -
Henrik Johansen -
Stéphane Ducasse