sources and scripts for generated files

sources and scripts for generated files

[Glenn Morris]

Does emacs-24 contain all the sources and scripts that are needed
to create any generated files in the tree?

E.g. do we need to add
http://www.unicode.org/ivd/data/2012-03-02/IVD_Sequences.txt
to admin/mac? It's ~ 1MB, and is used to create src/macuvs.h.

Re: sources and scripts for generated files

[Jan Djärv]

Hello.

26 maj 2014 kl. 04:31 skrev Glenn Morris <rgm@gnu.org>:

> 
> Does emacs-24 contain all the sources and scripts that are needed
> to create any generated files in the tree?
> 
> E.g. do we need to add
> http://www.unicode.org/ivd/data/2012-03-02/IVD_Sequences.txt
> to admin/mac? It's ~ 1MB, and is used to create src/macuvs.h.

IMHO, this is more a legal issue than a practical issue.
Say we add the file to Emacs.  There is no point in re-generating src/macuvs.h unless the file changes, and then the one distribiuted with Emacs is not used anyway.

But someone with insight to the GPL have to answer this.

	Jan D.

Re: sources and scripts for generated files

[Andreas Röhler]

On 26.05.2014 07:33, Jan Djärv wrote:
> Hello.
>
> 26 maj 2014 kl. 04:31 skrev Glenn Morris <rgm@gnu.org>:
>
>>
>> Does emacs-24 contain all the sources and scripts that are needed
>> to create any generated files in the tree?
>>
>> E.g. do we need to add
>> http://www.unicode.org/ivd/data/2012-03-02/IVD_Sequences.txt
>> to admin/mac? It's ~ 1MB, and is used to create src/macuvs.h.
>
> IMHO, this is more a legal issue than a practical issue.
> Say we add the file to Emacs.  There is no point in re-generating src/macuvs.h unless the file changes, and then the one distribiuted with Emacs is not used anyway.
>
> But someone with insight to the GPL have to answer this.
>
> 	Jan D.
>
>
>

Hi Jan,

assume you are kidding.

If not kidding, I'm still wondering, resp. asking: please make the GPL a text every Emacs core developer may understand without undergoing expensive law-studies.

Not to hide my personal view: of course there is no legal question. GPL requires the sources of built executable to be shipped, but not the stuff mayby useful for writing 
the sources.

Andreas

Re: sources and scripts for generated files

[Jan D.]

Hi.

Andreas Röhler skrev 2014-05-26 10:32:
> On 26.05.2014 07:33, Jan Djärv wrote:
>> Hello.
>>
>> 26 maj 2014 kl. 04:31 skrev Glenn Morris <rgm@gnu.org>:
>>
>>>
>>> Does emacs-24 contain all the sources and scripts that are needed
>>> to create any generated files in the tree?
>>>
>>> E.g. do we need to add
>>> http://www.unicode.org/ivd/data/2012-03-02/IVD_Sequences.txt
>>> to admin/mac? It's ~ 1MB, and is used to create src/macuvs.h.
>>
>> IMHO, this is more a legal issue than a practical issue.
>> Say we add the file to Emacs.  There is no point in re-generating
>> src/macuvs.h unless the file changes, and then the one distribiuted
>> with Emacs is not used anyway.
>>
>> But someone with insight to the GPL have to answer this.
>>
>>     Jan D.
>>
>>
>>
>
> Hi Jan,
>
> assume you are kidding.

No, I'm not.

>
> If not kidding, I'm still wondering, resp. asking: please make the GPL a
> text every Emacs core developer may understand without undergoing
> expensive law-studies.
>

The GPL says:

"The “source code” for a work means the preferred form of the work for 
making modifications to it."

Is macuvs.h source code?  I would say so.  Is it the form we prefer for 
modifications?  Maybe, maybe not.  But then again, why would we make 
modifications to the underlying Unicode document?  We can make 
modifications to the code that generates macuvs.h, thus we are able to 
make modifications in the "preferred form" IMHO.

You think all people agree in the interpretation of this case, simple as 
it may be?  Then you are kidding yourself.  That is why this is a legal 
question.


> Not to hide my personal view: of course there is no legal question. GPL
> requires the sources of built executable to be shipped, but not the
> stuff mayby useful for writing the sources.

First you say there is no legal question.  Then you cite a legal 
document making a legal requirement, thus underlying the fact that this 
*IS* a legal question.  Note that GPL does nowhere mentions "writing 
sources", it just mentions preferred form.  So your view is just an 
interpretation for which I can find no basis for in the GPL.

That does not say it is wrong, just that it is not clear-cut.

	Jan D.

Re: sources and scripts for generated files

[Stephen J. Turnbull]

Andreas Röhler writes:

 > assume you are kidding.

No, he's not.  GPL "sources" are defined as the preferred form for
editing if you wish to make changes to the software.  If the .h file
is as easy to edit as the other, it might be the preferred form for
local changes (or at least acceptable).  On the other hand, it's
converted by an automatic process *and* the script that does so is
part of Emacs (AIUI).  So there's a strong presumption that the file
from ICU is the source.

 > If not kidding, I'm still wondering, resp. asking: please make the
 > GPL a text every Emacs core developer may understand without
 > undergoing expensive law-studies.

That cannot be done; the GPL's primary purpose is to protect various
interests in the covered Work, and readability to non-lawyers takes a
back seat to that.

But this part is not hard to understand.

(1) You must provide sources that are reasonably convenient to edit.
(2) Which do you think is more editable?  Ship it.

Re: sources and scripts for generated files

[David Kastrup]

"Stephen J. Turnbull" <stephen@xemacs.org> writes:

> Andreas Röhler writes:
>
>  > assume you are kidding.
>
> No, he's not.  GPL "sources" are defined as the preferred form for
> editing if you wish to make changes to the software.  If the .h file
> is as easy to edit as the other, it might be the preferred form for
> local changes (or at least acceptable).  On the other hand, it's
> converted by an automatic process *and* the script that does so is
> part of Emacs (AIUI).  So there's a strong presumption that the file
> from ICU is the source.
>
>  > If not kidding, I'm still wondering, resp. asking: please make the
>  > GPL a text every Emacs core developer may understand without
>  > undergoing expensive law-studies.
>
> That cannot be done; the GPL's primary purpose is to protect various
> interests in the covered Work, and readability to non-lawyers takes a
> back seat to that.
>
> But this part is not hard to understand.
>
> (1) You must provide sources that are reasonably convenient to edit.

No, the preferred form for modification.  If there is an upstream
document from which the sources are generated _and_ this upstream
document is not one-of-its-kind (meaning that there will never be
updated documents in the same format), then obviously the upstream is
the preferred form for modifications.  Because then rerunning the
harvesting script is a proper part of maintenance, and then changes to
the generated document would get overwritten.

-- 
David Kastrup

Re: sources and scripts for generated files

[Andreas Röhler]

On 26.05.2014 11:23, Stephen J. Turnbull wrote:
> Andreas Röhler writes:
>
>   > assume you are kidding.
>
> No, he's not.  GPL "sources" are defined as the preferred form for
> editing if you wish to make changes to the software.  If the .h file
> is as easy to edit as the other, it might be the preferred form for
> local changes (or at least acceptable).  On the other hand, it's
> converted by an automatic process *and* the script that does so is
> part of Emacs (AIUI).  So there's a strong presumption that the file
> from ICU is the source.
>
>   > If not kidding, I'm still wondering, resp. asking: please make the
>   > GPL a text every Emacs core developer may understand without
>   > undergoing expensive law-studies.
>
> That cannot be done; the GPL's primary purpose is to protect various
> interests in the covered Work, and readability to non-lawyers takes a
> back seat to that.
>
> But this part is not hard to understand.
>
> (1) You must provide sources that are reasonably convenient to edit.
> (2) Which do you think is more editable?  Ship it.
>
>

Must confess that point of view is rather new for me :)

Preferences might change and be close to tastes. Should really all the ephemeric helper-functions used in the edit-process need to be included?

Always understood GPL legal term "source" in sense of required-to-build-the-binary.
I.e. in sense of "Corresponding Source", where it's said:

   The "Corresponding Source" for a work in object code form means all
the source code needed to generate, install, and (for an executable
work) run the object code and to modify the work, including scripts to
control those activities.

;;;;;;;;;;

I.e. a script needed for checking the dependecies would be a required part.
Some Emacs-script for editing-convenience not - otherswise any Emacs-edited GPL'ed must come with all the Emacs-stuff alongside, etc. That would point into eternity.

Cheers,

Andreas

Re: sources and scripts for generated files

[Stephen J. Turnbull]

Andreas Röhler writes:

 > > But this part is not hard to understand.
 > >
 > > (1) You must provide sources that are reasonably convenient to edit.
 > > (2) Which do you think is more editable?  Ship it.

 > Preferences might change and be close to tastes. Should really all
 > the ephemeric helper-functions used in the edit-process need to be
 > included?

Eh?

 > Always understood GPL legal term "source" in sense of
 > required-to-build-the-binary.  I.e. in sense of "Corresponding
 > Source", where it's said:

Before the definition of "Corresponding Source", the gods have chosen
to place the definition of "Source".[1]  See line one of the section
you quoted.  Application of that definition is what everybody else in
this thread is concerned with.

Footnotes: 
[1]  With apologies to Grandmaster Lasker, I believe it was, as well
as any atheists who may be listening in.

Re: sources and scripts for generated files

[Paul Eggert]

Glenn Morris wrote:

> Does emacs-24 contain all the sources and scripts that are needed
> to create any generated files in the tree?

It should contain a complete recipe, yes.

> E.g. do we need to add
> http://www.unicode.org/ivd/data/2012-03-02/IVD_Sequences.txt
> to admin/mac? It's ~ 1MB, and is used to create src/macuvs.h.

Yes, this should be like creating lisp/international/charprop.el from 
admin/unidata/UnicodeData.txt.  I gave it a shot in emacs-24 bzr 117157.

I notice a new version of IVD_Sequences.txt was released May 16.
We should upgrade.  Should we do this only in the trunk, or in the 
emacs-24 branch too?

Re: sources and scripts for generated files

[Stefan Monnier]

> E.g. do we need to add
> http://www.unicode.org/ivd/data/2012-03-02/IVD_Sequences.txt
> to admin/mac? It's ~ 1MB, and is used to create src/macuvs.h.

I don't think it's crucial, but it would probably feel at home under
admin/unicode.


        Stefan