Re: Cocoa texts and conversions?

Hi Piotr,

That is partly why I suggested it should be assigned to me (since 
I know the original code).  I agree that Council has *more* than 
enough work keeping up with maintenance of the Guidelines, as 
well as the Stylesheets and other software. This would be very 
low priority and if I didn't have time to do it in the next 
couple releases I'd note as much on the issue, and either I or 
someone could remove the cocoa-to-xsl scripts.  Though I'm 
curious what should be done when we do so.. previously we've just 
moved things around as and when we please or got rid of 
stylesheets , and that is good, but it might be beneficial to 
leave a note as to where these scripts could be found. (i.e. I 
might fork just the cocoa-to-xsl and stick it in my repositories 
in case it is useful to anyone, including me, in the future.) But 
then do we have to maintain that pointer? Maybe best just to 
delete and leave the existence of the closed issue on github as 
documentation. ;-)

But honest, I'll give it a good 20 minutes of thought in trying 
to fix it --  if I can't in that time, then maybe we'll get rid 
of it.

-James


On 10/01/17 12:26, Piotr Banski wrote:
> Hi James,
>
> I think the only good (or at least acceptable) argument is in 
> your last line (plus your request to get this assigned):
>
> > since it used to work I'm suspecting that making its output 
> valid again wouldn't necessarily be too difficult
>
> -- otherwise, I can't see why the Council should devote any 
> time to up-conversion from COCOA or any legacy format. This 
> should always, IMO, be an issue for a local project (formal or 
> informal: funded or performed as e.g. a semester assignment or 
> on a friend-to-friend basis). I feel scared by the prospect 
> that Council time (including F2F time) might get allocated to 
> such matters, seeing the flood of _current_ issues at GitHub.
>
> Best regards,
>
>   Piotr
>
>
>
>
> On 01/09/2017 12:57 PM, James Cummings wrote:
>> Hi Martin,
>>
>> Apologies for the delay in answering.  In case anyone is 
>> interested in their history: Yes, that initial cocoa-to-xsl 
>> script was indeed written by me (and then improved 
>> substantially into doing multiple passes by Sebastian with 
>> additional debugging output).  My first attempts at doing so 
>> were to make the Cocoa file well-formed XML (through wrapping 
>> a root element around it, changing any individual tags to be 
>> milestone-like psuedo elements) which was done in Perl before 
>> using XSLT, and then running that through two stylesheets, one 
>> to transform the flat cocoa markup language to a flattened TEI 
>> markup language (i.e. rename things) and then the next to fill 
>> that flattened TEI with structure (using xsl:for-each-group).  
>> The original files were written in 2004, with the express 
>> purpose of being used for a subset of the OTA consisting 
>> specifically of verse drama and I even spoke about this at 
>> ALLC-ACH 2004 in Gothenburg. The are available for now at 
>> http://users.ox.ac.uk/~jamesc/research/cocoa2tei/ where I put 
>> them in early 2006. I was doing this in a very modular way 
>> with the intent of applying to different sorts of texts as 
>> well. Partly this was created to estimate for a (failed) 
>> funding bid by the OTA how much work it would take to convert 
>> a large portion of the OTA archives all to TEI (whilst 
>> maintaining the original files of course).  [If I remember the 
>> reviews correctly it failed partly because one reviewer was 
>> outraged that mere technicians might *change* the files 
>> lovingly crafted by academics, even though the bid clearly 
>> spelled out that this was only format conversion and that the 
>> original files would also remain available.  Ah well.] In 2010 
>> as part of a business as usual drive to migrate some of the 
>> OTA texts into TEI P5, my email archive shows that Sebastian 
>> started working on a new cocoa-to-tei conversion and took my 
>> XSLT as the starting point and added the regex and an extra 
>> passes to do it all in a single XSLT while also making it a 
>> bit more general.  I contributed some bits and pieces.  That 
>> is probably why it isn't as generalised as it could be.  But I 
>> think it serves better as an example than a generalised 
>> conversion.
>>
>> Although the OTA is now in the Bodleian Libraries, not here at 
>> IT Services, I doubt it will ever get rid of its Cocoa texts 
>> (even if it provides migrated versions of them). While it may 
>> be easier to just get rid of the conversion I would certainly 
>> be willing to give it a go in trying to fix it. (i.e. If 
>> you've made a github issue, feel free to assign to me.) As to 
>> why we should have and maintain legacy conversions like this, 
>> I would say that it is as an example for those approaching 
>> similar migration up-conversions.  I'm not saying that is a 
>> very good reason, but since it used to work I'm suspecting 
>> that making its output valid again wouldn't necessarily be too 
>> difficult.
>>
>> -James
>>
>> On 07/01/17 18:40, Martin Holmes wrote:
>>> What I'm actually trying to do is understand the situation 
>>> better, and I'm getting there. This is what I think we've 
>>> learned:
>>>
>>>  - Cocoa is not very amenable to a generic conversion 
>>> process, since its syntax and usage is extremely variable;
>>>
>>>  - Our current Cocoa conversion script (James will hopefully 
>>> confirm or refute this) was written specifically to convert 
>>> certain OTA texts, and could not be applied generically;
>>>
>>>  - In any case, that script is broken (it produces invalid TEI);
>>>
>>>  - There is a possible way forward in writing a conversion 
>>> that creates milestone elements, as suggested by Lou, but 
>>> that would still depend on knowledge of the actual syntax and 
>>> special characters used in any specific set of texts;
>>>
>>>  - There are still some Cocoa texts out there which have not 
>>> yet been converted to TEI.
>>>
>>> What I take from this is that it would probably make more 
>>> sense to work with those people such as Elizabeth who have 
>>> Cocoa texts that matter, and get those converted to TEI; 
>>> meanwhile we could retire the conversion that doesn't work, 
>>> and remove it from the Stylesheets repo. This is obviously a 
>>> question for Council, though.
>>>
>>> Cheers,
>>> Martin
>>>
>>>>
>>>> cheers
>>>> stuart
>>>>
>>
>>
>


-- 
Dr James Cummings, [email protected]
Academic IT Services, University of Oxford