GEDCOM X

[Philip Durbin]

I just noticed http://en.wikipedia.org/wiki/GEDCOM#GEDCOM_X which led me
to http://gedcomx.org and https://github.com/FamilySearch/gedcomx

Forgive me if this has already been discussed on this list, but I'd be
curious to know if anyone has looked into GEDCOM X already.

The Wikipedia pages describes it as "a major new project around
genealogical standards".

Phil

--
http://greptilian.com

[Ron Savage]

Hi Philip

On 30/12/12 13:46, Philip Durbin wrote:
> I just noticed http://en.wikipedia.org/wiki/GEDCOM#GEDCOM_X which led me
> to http://gedcomx.org and https://github.com/FamilySearch/gedcomx
>
> Forgive me if this has already been discussed on this list, but I'd be
> curious to know if anyone has looked into GEDCOM X already.

I've come across it but I thought it had long been abandoned. I see
githubh says recent work on the code has taken place.

> The Wikipedia pages describes it as "a major new project around
> genealogical standards".
>
> Phil
>

--

Ron Savage
http://savage.net.au/
Ph: 0421 920 622

[Mikkel Eide Eriksen]

On 30/12/2012, at 05.08, Ron Savage <r...@savage.net.au> wrote:
> On 30/12/12 13:46, Philip Durbin wrote:
>> I just noticed http://en.wikipedia.org/wiki/GEDCOM#GEDCOM_X which led me
>> to http://gedcomx.org and https://github.com/FamilySearch/gedcomx
>>
>> Forgive me if this has already been discussed on this list, but I'd be
>> curious to know if anyone has looked into GEDCOM X already.
>
> I've come across it but I thought it had long been abandoned. I see githubh says recent work on the code has taken place.
>
>> The Wikipedia pages describes it as "a major new project around
>> genealogical standards".

Hi Ron, Phil

I've been following it somewhat on & off for a few months. It's still very much a work in progress, with drastic changes being made at times. It is my understanding that FamilySearch are talking privately with a number of other commercial genealogy software/data providers before making decisions. However, they're open to discussion on the Github issue tracker, and a couple of my suggestions have been accepted.

Recently, much work was done on the Place model -- see https://github.com/FamilySearch/gedcomx/pull/79 -- and current work appears to be concentrated on multimedia links & attachments -- see https://github.com/FamilySearch/gedcomx/pull/224

One thing I like about the idea of GedcomX is the emphasis on sourcing conclusions, which will make the workflow more documentation-centric (most Gedcom-based apps I've tried treat sources almost as an afterthought). Hopefully a useful standard will emerge in the end.

I suggest taking a look at their blog on gedcomx.org and the specifications:

https://github.com/FamilySearch/gedcomx/tree/master/specifications

Especially the conceptual model.

Regards,
Mikkel Eide Eriksen (@mikkelee on github)

[Paul Johnson]

On Sun, Dec 30, 2012 at 10:13:53AM +0100, Mikkel Eide Eriksen wrote:
> On 30/12/2012, at 05.08, Ron Savage <r...@savage.net.au> wrote:
> > On 30/12/12 13:46, Philip Durbin wrote:
> >> I just noticed http://en.wikipedia.org/wiki/GEDCOM#GEDCOM_X which led me
> >> to http://gedcomx.org and https://github.com/FamilySearch/gedcomx
> >

> > I've come across it but I thought it had long been abandoned. I see
> > githubh says recent work on the code has taken place.
>

> I've been following it somewhat on & off for a few months. It's still
> very much a work in progress, with drastic changes being made at
> times.

I was following the effort until github changed their notifications a
few months ago. I must admit to being not particularly impressed.

GEDCOMX seemed to arrive full of bluster, Java and XML, and gave the
impression that the only unsolved problems were things like deciding how
to store the XML and then just finishing the programming. That was
rather uninteresting to me.

Far more interesting was the data model, and despite the programming
going on, GEDCOMX didn't seem to have that nailed down. And as far as
it was exposed, it didn't really seem to build on any of the other work
that had taken place in this area over the last decade or so.

Other people also noticed this and so there was some measure of working
on specifications, but whilst the wiki at
https://github.com/FamilySearch/gedcomx/wiki does describe the
high-level aims of the project it's still unclear to me exactly how
GEDCOMX is planning to solve those problems. As you can see from that
page, the link to the data model points to a page which has yet to be
written.

Perfect is the enemy of good, and this has dogged many similar
initiatives, but GEDCOMX seems to have swung in the opposite direction
entirely. I would be pleased to see the project make real progress but
I'm afraid that I find it unlikely.

--
Paul Johnson - pa...@pjcj.net
http://www.pjcj.net

[Ron Savage]

Hi Mikkel

Thanx for the extra info.

On 30/12/12 20:13, Mikkel Eide Eriksen wrote:
> On 30/12/2012, at 05.08, Ron Savage<r...@savage.net.au> wrote:
>> On 30/12/12 13:46, Philip Durbin wrote:
>>> I just noticed http://en.wikipedia.org/wiki/GEDCOM#GEDCOM_X which led me
>>> to http://gedcomx.org and https://github.com/FamilySearch/gedcomx
>>>
>>> Forgive me if this has already been discussed on this list, but I'd be
>>> curious to know if anyone has looked into GEDCOM X already.
>>
>> I've come across it but I thought it had long been abandoned. I see githubh says recent work on the code has taken place.
>>
>>> The Wikipedia pages describes it as "a major new project around
>>> genealogical standards".
>
>
> Hi Ron, Phil
>

> I've been following it somewhat on& off for a few months. It's still very much a work in progress, with drastic changes being made at times. It is my understanding that FamilySearch are talking privately with a number of other commercial genealogy software/data providers before making decisions. However, they're open to discussion on the Github issue tracker, and a couple of my suggestions have been accepted.
>
> Recently, much work was done on the Place model -- see https://github.com/FamilySearch/gedcomx/pull/79 -- and current work appears to be concentrated on multimedia links& attachments -- see https://github.com/FamilySearch/gedcomx/pull/224

>
> One thing I like about the idea of GedcomX is the emphasis on sourcing conclusions, which will make the workflow more documentation-centric (most Gedcom-based apps I've tried treat sources almost as an afterthought). Hopefully a useful standard will emerge in the end.
>
> I suggest taking a look at their blog on gedcomx.org and the specifications:
>
> https://github.com/FamilySearch/gedcomx/tree/master/specifications
>
> Especially the conceptual model.
>
> Regards,
> Mikkel Eide Eriksen (@mikkelee on github)

[Mikkel Eide Eriksen]

On 30/12/2012, at 22.19, Paul Johnson <pa...@pjcj.net> wrote:
> Perfect is the enemy of good, and this has dogged many similar
> initiatives, but GEDCOMX seems to have swung in the opposite direction
> entirely. I would be pleased to see the project make real progress but
> I'm afraid that I find it unlikely.

I agree, there are many unsolved issues with GedcomX, and the specification is nowhere near mature enough for anyone to have started writing actual code -- but I figure it'll be slightly more likely to get to that point if I contribute when I can.

Mikkel

[Paul Johnson]

I absolutely agree, and I applaud you and everyone else who is doing so
(thanks for opening the data model ticket). I hope that at some point
the project will get to a point where someone can make an implementation
in Perl.

[Ron Savage]

Hi Mikkel

One thing I noticed is the very inadequate attempt to specify a street
address.

Many years ago I ran a mailing list to print sticky labels for a little
magazine. I live in Melbourne Australia and a recipient lived inside a
hospital in London. His street address component required 6 lines,
whereas GedcomX only specifies street/street2/street3 :-(.

[Ron Savage]

Hi Paul

On 31/12/12 08:19, Paul Johnson wrote:
> On Sun, Dec 30, 2012 at 10:13:53AM +0100, Mikkel Eide Eriksen wrote:
>> On 30/12/2012, at 05.08, Ron Savage<r...@savage.net.au> wrote:
>>> On 30/12/12 13:46, Philip Durbin wrote:

> GEDCOMX seemed to arrive full of bluster, Java and XML, and gave the
> impression that the only unsolved problems were things like deciding how
> to store the XML and then just finishing the programming. That was
> rather uninteresting to me.
>
> Far more interesting was the data model, and despite the programming
> going on, GEDCOMX didn't seem to have that nailed down. And as far as
> it was exposed, it didn't really seem to build on any of the other work
> that had taken place in this area over the last decade or so.

I'm impressed by the amount of work which has been done, and a bit
disconcerted where it ties in to Java. I do see the attempt to adopt
material from the Java world, and elsewhere, but the link to Java seems
a bit too tight for my liking.

Likewise adopting the Zip format, as its max size limitations become
more and more apparent, given their comment about (optionally) storing
everything in a single file (if I understood that, of course).

Overall though, I'm worried if they haven't over-specified things, in
classic Java style, which leaves me feeling there may have to be a
auxiliary spec answering the question: What are they really saying (at
each point)? This would collapse several layers of definitions, if needed.

[Stephen Woodbridge]

I spent some hours reading through the specs and I come away with the
follow thoughts:

1. I like what they are trying to do
2. I think it is overly complex, but may be it needs to be
3. There are major holes or areas that things are under specified, for
example:
* addresses as mentioned
* place names as a single string that would require additional parsing
to make sense of
* events a lot of defined but a lot are not, like military induction for one
4. I didn't like the connection to Java either, I see no need for that
in the spec, if there is some point that they are trying to establish by
it, it would be better to just spell it out in details so people that do
not work with Java have direct guidance.

And probably other things. I think it would require a lot of UI stuff to
hide the complexity of this but that should not be a show stopper.

-Steve

[Philip Durbin]

On 12/30/2012 07:05 PM, Stephen Woodbridge wrote:
> 4. I didn't like the connection to Java either, I see no need for that
> in the spec, if there is some point that they are trying to establish by
> it, it would be better to just spell it out in details so people that do
> not work with Java have direct guidance.

Right, I thought it was odd to read "the GEDCOM X data model is defined
in Java" at http://www.gedcomx.org/Code.html

Likewise, http://www.gedcomx.org/Community.html says "The GEDCOM X
schemas and interfaces are defined in Java."

Thank you, everyone, for the great discussion on GEDCOM X. It seems like
something to keep an eye on.

Phil

[Mikkel Eide Eriksen]

I think I saw one of the devs (possibly Ryan Heaton @stoicflame) comment on an issue at some point that they were moving away from defining it in terms of Java and rather to have a language-agnostic spec with a reference implementation in Java. I can't find it just now, but in any case, I should think that would be the correct way of dealing with it. And if I'm not misremembering, that should be put on their site in big fat letters.

Perhaps I'll open a ticket next year ;) if I don't find where they said it.

Happy new years,
Mikkel

[Ron Savage]

Hi Mikkel

Yes, open a ticket, politely but firmly reminding them that a
Java-centric view of the world is not the only possible view...

Happy New Year!

And here's hoping this year is even more productive that the last, which
for me at least, was very productive :-).

[Roland Talanow, M.D., Ph.D.]

Please unsubscribe me finally!
I unsubscribed now three times and after a few months am back on this email
list.
I like Gedcom but not participating anymore - so PLEASE have mercy!
Thank you and have a Happy New Year,
Roland

Mikkel=

[Mikkel Eide Eriksen]

Hi all,

Took a little longer than expected due to other commitments, but I've created an issue here:
https://github.com/FamilySearch/gedcomx/issues/228

Please comment as you see fit.

Mikkel

[Philip Durbin]

No, it took a long time because it's so thoughtful and thorough! Great job,
Mikkel!
On Jan 19, 2013 3:53 AM, "Mikkel Eide Eriksen" <mikkel....@gmail.com>
wrote:

[Ron Savage]

Hi Mikkel

Yes, well done. Firm, clear, without being strident.

[Mikkel Eriksen]

Hi,

An update: FamilySearch has now moved to accomodate our objections.

- Java has been excised from the gedcomx repository and now resides in
a separate repo.
- They are aware of the documentation shortcomings, and are working to
address them.

In addition, tonight a number of new proposals showed up that look
very interesting, especially regarding the "persona" construct which I
think could be a very good solution to the problem of conflicting
sources. I will continue to try and push them in a direction I feel
would support my needs, and I encourage you to do the same :)

Regards,
Mikkel

PS. I'll try not to "spam" this list about GedcomX any more!

[Philip Durbin]

I only just saw this...

Great news and good work!

[Ron Savage]

Hi

They also accepted my suggestion to increase the # of street address
lines (from 2 IIRC) to 6. I explained I needed that many once for
someone who lived in a hospital in London.