[sword-devel] Bibles in Beta

[DM Smith]

Peter von Kaehne wrote:
> The way we work with software is
>
> 1) multiple independent projects
> 2) each project has one or several leads
> 3) there are other contributors with and without write access to the
> project repository
> 4) all projects are in some form of version control
> 5) Many projects have a development version and (in)frequent releases
> 6) we work towards three aims a) bug freeness b) OSIS feature
> completeness and c) new and imaginative software features
>
> Much of the above could be fully applicable on our work with modules -
> but we do not do this. Why?
>
> If we would create a way of
>
> a) having a raw text repository where the texts are kept in some import
> format (OSIS, ThML, whatever) under version control
>   
I think this needs to be the original text as we received it. Not the 
import format. We should keep complete records of when, where, why and 
how we got the text. Because the "place" where we got it might change 
(the source is updated, or no longer available, ....), I don't think 
that merely noting where it was obtained is sufficient.

If I had the original text, I may have been able to compare the 
resulting modules in beta for whether the "missing" verses were missing 
in the source.

Along with this should be the script(s) used to create the import format 
to our mod creation tools, along with sufficient instruction for anyone 
(with permission) to be able to repeat the transformation.

I've suggested a repository of source in the past, but Troy was 
reluctant/against such a thing. IIRC, his desire was that we did not 
become a secondary source for texts, and that modules as an opaque 
representation would be the only publication of texts by CrossWire 
(excepting "original" work such as the KJV). (Troy, if I mis-represented 
you, please speak up:) I don't know if having it under lock and key, 
with a select group having keys, would satisfy his concerns.

> b) with named module owner(s) and contributor(s) -
> c) with some form of automated build mechanism which could create the
> modules in all the various formats we require (dmg, winzip, zip and raw)
>   
I think this is automated by the module download page. Were you thinking 
of something more? (BTW, dmg is not possible without a Mac.)

> d) have a discussion what constitutes "good enough" for a release and
> what is good enough for a beta + implement two finished module
> repositories - "public" and "beta" similar to now, but let the module
> owners decide when they are "there"
>   
To me, good enough includes:
Some level of transformational accuracy.
Some level of completeness in the conf that goes beyond what is 
minimally necessary for the module to work. This often is a user's first 
impression of the module.

> e) work with each module towards increasing compliance with a) textual
> correctness/completeness and b) maximalised features
>
>  The raw text  repository could be world readable for all PD modules,
> but for others it would be read-restricted to crosswire contributors (or
> even just module owner)
>
> This would be a major shift for us, but I think it would be the way to
> fully harness our ability to work in groups, it would be much closer to
> our ways as volunteers working in starts and spurts, it would improve
> drastically our ability to publish without dumbing down our output -
> rather the opposite.
>
> Peter
>   
>
>
> Peter von Kaehne wrote:
>   
>> The TurNTB has often verses grouped together, but this is not encoded
>> ie.g. as  v4-6 but instead v4 has all content and v5 and v6 are empty.
>> So the list you showed for it is endless in consequence.
>>
>> I would think this + alternative versification is the case in many other
>> Bibles.
>>
>> Personally I would suggest we are shooting us in our feet if we insist
>> on perfection. "Good enough" is a hard to define but crucial standard.
>>
>> My suggestion is to put this list in a suitable format into each module
>> as a list for further checking/errata and publish them nevertheless. If
>> whole books or whole chapters are missing they need to be listed in the
>> Conf file.
>>
>> With all these modules we should be working in an eternal (but hopefully
>> ever slowing down) cycle of continuous improvement rather than
>> delivering perfection once and then being able to forget about it.
>>
>> The real problem is that the structure of our repositories and the
>> current access rules make it extremely hard to work in such a fashion.
>> We need to find new ways of working on the modules, which are akin to
>> the way we work on software - gradual improvements, frequent releases
>> and many contributors.
>>
>> My suggestion - and I am talking here without having this fully though
>> through:
>>
>> 1) lets clear out Beta now rapidly and divide the modules into actually
>> working and actually not working + release the lot - even if there are
>> flaws. Include in the released one an errata file for now
>>
>> 2) have a broad and thorough discussion on how we are going to run the
>> module repository subsequently in order to facilitate increased content,
>>  constantly improving content in terms of increasing features and
>> decreasing error
>>
>> Peter
>>
>> DM Smith wrote:
>>     
>>> For what it's worth, I've "scoped out" the Bibles in beta, comparing 
>>> them to the KJV versification. (I skipped the GenBook Bibles) I wrote a 
>>> little program using JSword to do the work. Hopefully, I didn't make a 
>>> mistake in writing it. (I've spot checked it against GnomeSword.)
>>>
>>> Below are the results. They are broken up by testament (i.e. OT and NT) 
>>> and by verses in scope and verses not in scope (i.e. InScope, OutScope). 
>>> Obviously, only one of these was necessary, but it is quicker to note 
>>> potential problems. The results are displayed as osisRefs.
>>>
>>> I find some of the results surprising:
>>> The last verse of many chapters is missing.
>>> Malachi 4 is missing from many Bibles.
>>>
>>> While this may be just an alternate versification, I think that it might 
>>> be a good idea to spot check the "missing" verses against the source 
>>> (where ever it came from) and possibly a hard copy. If anyone can pitch 
>>> in that would be great.
>>>
>>> My personal feeling is that we should not release incomplete Bibles. 
>>> (like turNTB!!!)
>>>
>>> In Him,
>>>     DM
>>>       
.... listing deleted ....