[Cuis-dev] ODB options
H. Hirzel
hannes.hirzel at gmail.com
Tue Dec 17 00:00:00 PST 2024
Hello
I think it is good to look for different persistence solutions at least for
2025.
On Sat, Dec 14, 2024 at 7:24 PM Hilaire Fernandes via Cuis-dev <
cuis-dev at lists.cuis.st> wrote:
> The educator model[1] is straightforward to persist with reference stream
>
But it could also just just a number of CSV files. A dynabook app part
managing planning schedules, task and time tables well is an app in its own
right which is will be welcomed by teachers. There are situations where
each student has his or her own schedule. Also general information is
available in table format [3] A user will have more trust in the app if he
or she sees that the information is stored in a directory with
subdirectories with CSV data
> but it is more complicated with the knowledge model[2] where each document
> should be saved in its own refstream file and loaded at demand, otherwise
> the whole model (educator+knowledge) may become too big and slow to
> save/load. Imagine the activity of a teacher or a learner along one school
> year, a lot of documents will be produced.
>
I understand that teachers and students produce dynabook documents and
exchange them. That means that a format for exchanging these documents is
needed. Probably a binary one but not necessarily. Or a combination of it.
MS Powerpoint and LOImpress also have pages/sildes and use various data
formats. This format then is also the one to be used for persistence.
> I play a bit with DiskProxy so that document referenced in the educator
> model are saved as that, but I suspect it ends all the document model are
> still load at once. Document are a mix of view, model and presenter so it
> makes a lot, and it is not easy to control how far in the tree of objects
> the persistence goes, then other problems arise at realod, like
> disconnected object.
>
Would it be possible to de-couple persistence needs?
Do the views need to be persisted? See recent discussion on this list about
recreating morphs from source code, i.e. getting the source code for a
morph tree and then using it to rebuild it.
I continue my answer by replying to the mail Juan wrote.
--Hannes
>
> [1]
> https://github.com/Dynamic-Book/doc/tree/main/technical/DyboApp#educator-objects
>
> [2]
> https://github.com/Dynamic-Book/doc/tree/main/technical/DyboApp#knowledge-objects
> Le 14/12/2024 à 15:58, H. Hirzel via Cuis-dev a écrit :
>
> Could you please elaborate about your mixed experience with
> ReferenceStream? To know that would be helpful.
>
>
[3] As an illustration for the need of CSV data and that it makes sense to
actually show the data in a directory structure: For example school
holidays:
France:
https://www.service-public.fr/particuliers/vosdroits/F31952?lang=en
Switzerland
https://publicholidays.ch/school-holidays/
Germany
https://publicholidays.de/school-holidays/
Italy
https://publicholidays.it/school-holidays/
The dynabook 'Cash register type' machine could start out with a well
elaborated planning/scheduling part prepackaged with information needed
for this.
> -- http://mamot.fr/@drgeo
>
> --
> Cuis-dev mailing list
> Cuis-dev at lists.cuis.st
> https://lists.cuis.st/mailman/listinfo/cuis-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cuis.st/mailman/archives/cuis-dev/attachments/20241217/c913a054/attachment.htm>
More information about the Cuis-dev
mailing list