[squeak-dev] Fully Relicensed version of mainstream Squeak

  Hello,

  First of all, thank you for all the people who spend great deal of
time on fixing our server.  I've been watching the IRC log
(http://tunes.org/~nef/logs/squeak/) and was admiring the work!

  As mentioned in the leadership discussion minutes from Craig, the
plan now is as follows:

  - Make Squeak version 4.0.  This is based on the 3.11 effort but
    get rid of or rewrite code that are not relicensed and make a
    fully relicensed version relatively conservatively.  Etoys 4.0 is
    now fully relicensed, and we can bring the removal and rewrite
    changesets from that stream.

  - Craig continue to work on the Spoon based system.  It is dubbed
    Squeak 5.0.  (My personal opinion is that because it is fairly
    different, it could have a different name, but...)

BTW, during the Etoys' relicensing effort, I made a little web app
that lets you view *all history* from Squeak V1 to the latest version:

http://tinlizzie.org:8080/seaside/examples/authorship2

I can make a similar page for 3.10 or such, and also give a tool to
check the unlicened code in a particular code base.

  Ken and Mathew, how does it sound?

-- Yoshiki

On Wed, Oct 22, 2008 at 05:24:58PM -0700, Yoshiki Ohshima wrote:
Keith would be the interested person, not Ken.

It sounds like a good plan. Do you have a patch I could use to
get started? I would definitely like to see just how bad the
situation is in 3.9 or 3.10. Croquet would be nice too, but I
think the history of Croquet is much more muddled than the
history of squeak.org

--
Matthew Fulmer -- http://mtfulmer.wordpress.com/

> Keith would be the interested person, not Ken.

  Ah, sorry.

> It sounds like a good plan. Do you have a patch I could use to
> get started? I would definitely like to see just how bad the
> situation is in 3.9 or 3.10.

  Here is an announcement I made while ago:

http://lists.squeakfoundation.org/pipermail/squeak-dev/2008-September/131586.html

This is based on a bit old database, but you can get the feel of it.
Attached is the latest version of my Authorship2 code: you should be
able to file it in and see the current status.

> Croquet would be nice too,

  Nice in terms of what?

> but I think the history of Croquet is much more muddled than the
> history of squeak.org

  Well, what makes you think so?  The set of developers of the Croquet
image is definitely more limited, and the "team" generally had better
control on the image than the main stream.

-- Yoshiki

>   Here is an announcement I made while ago:
>
> http://lists.squeakfoundation.org/pipermail/squeak-dev/2008-September/131586.html
>
> This is based on a bit old database, but you can get the feel of it.
> Attached is the latest version of my Authorship2 code: you should be
> able to file it in and see the current status.

  I forgot to mention one thing.  The updates are at:

http://tinlizzie.org/updates/etoys/updates/

.  Take a loot at

http://tinlizzie.org/updates/etoys/updates/updates.list

and fetch the updates for "etoys3.2".

-- Yoshiki

Matthew Fulmer wrote:
> It sounds like a good plan. Do you have a patch I could use to
> get started? I would definitely like to see just how bad the
> situation is in 3.9 or 3.10. Croquet would be nice too, but I
> think the history of Croquet is much more muddled than the
> history of squeak.org

Probably not. Remember that both Tweak and Croquet were MIT licensed
from day one which really simplifies matters since all contributions
were always understood to be MIT licensed. The only question is whether
the Croqut base image (deployment/Homebase.image) is squeaky license
clean or not (pun intended). A quick check - comparing the versions in
homebase.image with the versions in etoys4 - shows that any auditing
effort should be *very* manageable (mostly because they share the same
3.8 ancestry). Here is the total of methods that are in Homebase.image
and where etoys4.image has different versions (scripts below):

Kernel:         245 / 3035 (91.92% identical)
Collections:    124 / 2047 (93.94% identical)
Compression:     13 / 435  (97.01% identical)
Multilingual:   179 / 1023 (82.5% identical)
Graphics:       110 / 2196 (94.99% identical)
GraphicsTools:  1 / 179  (99.44% identical)
TrueType:         9 / 140  (93.57% identical)
XML-Parser:      21 / 339  (93.8% identical)
Network:        195 / 1910 (89.79% identical)
Morphic:       1349 / 11534(88.3% identical)
SmaCC:            0 / 113  (100.0% identical)
Sound:           61 / 1523 (95.99% identical)
Files:           28 / 513  (94.54% identical)
System:         469 / 3317 (85.86% identical)
Exceptions:      13 / 229  (94.32% identical)
Balloon:         72 / 462  (84.41% identical)
Movies:          61 / 524  (88.35% identical)
Flash:           20 / 723  (97.23% identical)
Tools:          203 / 1946 (89.56% identical)
ST80:            50 / 1559 (96.79% identical)
Nebraska:       169 / 707  (76.09% identical)
FFI:             10 / 393  (97.45% identical)
Compiler:        94 / 657  (85.69% identical)
Protocols:       15 / 297  (94.94% identical)

SUnit:            0 / 221 (100.0% identical)
Tests:            1 / 394 (99.74% identical)
MorphicTests:     6 / 47  (87.23% identical)
NetworkTests:     0 / 48  (100.0% identical)
KernelTests:     13 / 657 (98.02% identical)
GraphicsTests:    0 / 79  (100.0% identical)
CollectionsTests: 0 / 115 (100.0% identical)

SMBase:         335 / 603  (44.44% identical)
SMLoader:        21 / 98   (78.57% identical)
Games:          613 / 1358 (54.86% identical)

PackageInfo-Base: 41 / 134 (69.4% identical)
Monticello:    1763 / 1763 (0.0% identical)
VersionNumber:    0 / 36   (100.0% identical)
TweakMC:         21 / 21  (0.0% identical)
MonticelloConfigurations: 113 / 113 (0.0% identical)

ToolBuilder-Kernel:  305 / 305 (0.0% identical)
ToolBuilder-Morphic: 142 / 142 (0.0% identical)
StarSqueak:     208 / 209 (0.47% identical)
Scamper:         90 / 90  (0.0% identical)
VMMaker:       3507 / 3531(0.67% identical)
Speech:         954 / 954 (0.0% identical)

Note that for most of the core packages (like Network or Morphic) I
would expect the changes to come from the etoys image due to the
modifications done for OLPC. In reality I would expect that you would
probably have to audit less than 5% of the code base. And of that I
would expect a goodly chunk to be written by yours truly which means it
is either covered by the VPRI or the Croquet license.

Cheers,
   - Andreas

[--------- Creates a file with all versions of all methods ----------]
"Run this in homebase.image"
Utilities informUserDuring:[:bar|
allVersions := MCWorkingCopy allManagers collect:[:mgr|
        info := mgr package packageInfo.
        bar value: 'Processing ', info name.
        Array with: info name with: (info methods collect:[:mref|
                Array with: mref actualClass name
                        with: mref methodSymbol
                        with: mref compiledMethod timeStamp]).
]].
(FileStream forceNewFileNamed: 'C:\homebase.versions')
        fileOutClass: nil andObject: allVersions;
        close.

[-------- Compares the above file with the current image ---------]

allVersions ifNil:[
        [file := FileStream readOnlyFileNamed: 'C:\homebase.versions'.
        allVersions := file fileInObjectAndCode] ensure:[file close].
].
Utilities informUserDuring:[:bar|
allVersions do:[:pspec|
        pkgName := pspec first.
        bar value: 'Verifying ', pkgName.
        total := matches := 0.
        pspec second do:[:mspec|
                cls := Smalltalk classNamed: mspec first.
                cls ifNotNil:[method := cls compiledMethodAt: mspec second ifAbsent:[]].
                method ifNotNil:[stamp := method timeStamp].
                stamp = mspec third ifTrue:[matches := matches + 1].
                total := total + 1.
        ].
        Transcript cr; show: pkgName,': ', (total-matches) asString,' / ',
total asString,' (', (matches asFloat / total * 100 truncateTo: 0.01)
asString, '% identical)'.
]].

On Oct 23, 2008, at 5:27 AM, Yoshiki Ohshima wrote:

>> Here is an announcement I made while ago:
>>
>> http://lists.squeakfoundation.org/pipermail/squeak-dev/2008-September/131586.html
>>
>> This is based on a bit old database, but you can get the feel of it.
>> Attached is the latest version of my Authorship2 code: you should be
>> able to file it in and see the current status.

I could not see the authorship2 code.
Did you send it?

At Thu, 23 Oct 2008 15:11:04 +0200,
stephane ducasse wrote:
>
> I could not see the authorship2 code.
> Did you send it?

  No, but I uploaded code to SqueakSource.

http://www.squeaksource.com/MethodAuthorship/MethodAuthorship-yo.1.mcz

is the data collection mechansim (I forgot to set the ancestor so the
version number of the mcz package is wrong) and the Seaside code is

http://www.squeaksource.com/AuthorshipViewer.html

-- Yoshiki