ORM-ish à la TiddlyWiki: Project Updates

[Charlie Veniot]

Project TiddlyWiki:  ORM-ish à la TiddlyWiki

BTW:  My announcement for this project has turned into an awesome thread of discussion.  Please check out Using TiddlyWiki for fact-based information modelling and database engineering ??? for some ridiculously excellent buffet of food for thought about, um, thought (cognitive processing).  Things like (I can't do it all justice):

• managing complexity
• information/knowledge design/navigation
• PhD-related stuff about context and keeping stuff understandable
• in short:  A LOT OF WOW GOODNESS !!!

Latest Updates

(The following pretty much sums up changes/additions since November 23rd.)

Text Rotation Testing

• I took a little side trip trying to figure out a generic mechanism from creating tables like in the "Entity Relationship Matrix" tiddler.  Although I gave up after a few hours (much too big and messy of a job), I left the following two tiddlers in the TiddlyWiki in case I ever want to try again: "Research Macros" and "Text Rotation Tester"
  

Coded Form of Facts

• I last had coded form of facts setup as one field for each fact, containing the entire coding line (for example:  coded field = "BUILDING 1:1 BLDG_NAME").
• That was making all of my scripting more complicated (having to split that coded field into the three parts: BUILDING, 1:1, and BLDG_NAME for whatever kind of processing.)  MORE IMPORTANTLY: that was making it much more difficult for me to setup "data entry" of a fact's coded form via select widgets.
• Now, a fact's coded form is split into three fields: c1, c2, and c3.  (for example:  c1 field = "BUILDING", c2 field = "1:1", and c3 field = "BLDG_NAME".)
• This change from one field to three required some rejigging of various tiddlers, including (among others):  "Entity Relationship Matrix", "oEs" (a transclusion template to generate conceptual representations of entities; see "BLDG_SPACE_UNIT" tiddler for an example)
  

Early start (experimentation) of widgets for data entry

• Using "select widgets" to help with creating coded form of facts.  I'm very new to the use of widgets for this purpose, so this will take me a while to get right.  For a starting example, see the "A Building has exactly one Name / A Name is unique to a Building" fact

Entity Relationship Matrix tiddler

• For the slanted column titles, I removed the little links for the related tiddlers, because the slanting of titles often causes the tiddler links to not be clickable (they wind up at the bottom of "z-order", and it just isn't a fun thing to try and fix at the moment.)
• Pure CSS for tooltips on the check marks is wonky sometimes because of tooltip positioning (which can only be auto-adjusted nicely to show the tooltip, in a guaranteed viewable spot, with javascript.)  That is not a fun thing to try and fix at the moment.
• Scrolling the table columns towards the left: modified so that when columns get slid over to the left, the cells don't show on top of the row headers on the far left (that remain in fixed position.)
  
  • I haven't figured out how to do the same for column headers, but I'm on the fence: not sure I want to hide column headers when columns are scrolled to the left.

Primary Keys for entities

• I was setting up all primary keys as "value_type" tiddlers, but I've modified so that primary keys are now specified as a simple field on each related entity tiddler.  
• Seeing as I'm a big believer in using Sequence Numbers (as per Oracle Database products; whatever is equivalent in other database products), it doesn't make much sense for each primary key having a dedicated tiddler.
  • Instead, every entity has a sequence number name specified in a field, and each sequence number follows the exact same standard (data type, max length .... all of which are currently hard-coded in scripts; I plan to move those "standard sequence number details" in a data tiddler).
• BTW: Related oEs template tiddler (for conceptualisation of entities) modified accordingly.
  

Conceptualisation of Entities

• For each attribute, the table now shows the related fact in a new column on the far right side.  (Just nice extra information about each attribute).
• TODO: setup some mechanism to specify an order for the attributes (likely up/down buttons with saving position in related fact tiddlers; still thinking about it.)
  

Task Tracking

• Renamed "Facts - Coded Form Pending" to "Facts - Incomplete Coded Form".
• Got rid of "Entities/Attributes Pending Creation" tracking because "Facts - Incomplete Coded Form" handles that.
• Added "Entities - Incomplete Primary Key" tracking.
  

Entity "... to Many" relationships with value types or with other entities

(getting into some heavy-ish relational database design stuff here ...)

• In simple "... to 1" relationships, this would translate into one table for the one entity (for an entity relationship to a value type), or a table each for the two entities (for an entity relationship to an entity).  Easy peasy.
• However, for a case of "... to many", that introduces an additional table for both scenarios (1. relationship to many value type instances 2. relationship to many entity instances.)
  • That's something I need to ponder on: i.e. how to handle writing facts, coding those facts, and conceptualisation entities (i.e. the related scripts) 

[Charlie Veniot]

"ORM-ish à la TiddlyWiki" is now "The Tifoist Project"

• Tifoist Is a Fact-Oriented Information-Semanticization Tool

The previous work I now call "prototype 1."

I am now labouring on "prototype 2", involving some GUI design and overhauling/refactoring what I did in prototype 1.  Much of what I did in prototype 1 is now broken in prototype 2.

Much work to do before prototype 2 can be released into the wild for some scrutiny.  For now, I share a few screenshots (attached). 

Cheers !

[stan...@gmail.com]

This looks really nice, Charlie.  Let me know when you are ready for some beta testing.  I have a couple of nice book projects where I could use Tifoist.

Stan

[Charlie Veniot]

For sure, Stan.  I've got to iron out some kinks, and figure out how a user could upgrade from one version of Tifoist to the next (i.e. never need to re-enter or fix anything in an upgrade.)

Once I can get the basic functionality working, I will throw myself into figuring out how to build a Tifoist plugin.

I'm thinking, very much on the fly:

• Testing of basic functionality first.  Iron kinks.
  • Well, once I finish "basic functionality", (the scope is a moving target right now.)
• Setup Tifoist as a plugin, then test that and iron out kinks.
• Then work on Tifoist upgrading.  Test and iron out kinks.
• At some point, test TiddlyWiki upgrade of a TiddlyWiki with Tifoist plugin.
• At some point, and in logical chunks, add more functionality, while making sure none of the above get broken.

I've spent the last few days hyperfocused on GUI design (i.e. making the TiddlyWiki look like an an application for domain "modeling" and database engineering, maybe even software engineering).

Also a little sidetracked by naming paralysis by analysis.

Anyhoo, latest incarnation for your viewing pleasure and scrutinizing:  Tifoist-TW5 Prototype 2 (continuing to model a domain I know well: the business domain of New Brunswick's Buildings Group, related information about 2/3rds of the way down this page.)

[Hans Wobbe]

cj.v:

I've been watching your ideas with considerable interest since I have access to a LOT of realEstate/property data and I am surrounded by a large group of SQL experts.  (For my part, it's been at least 20 years since I used SQL and, now,  I would/should be blocked from doing so again by the techs I work with).

I cannot help but wonder is there is enough synergy that we might be able to propose/build a bit of a prototype of how TW and SQL might interact (even if only at the Design stage).  I'd be content to explore this further using email either at HWo...@DataFix.com (Professional use - heavily spam filtered, so I might need to punch a hole for you) or HwW...@gMail.com (Personal use).

Cheers,

Hans

[Charlie Veniot]

G'day Hans,

That sounds wickedly interesting, but way beyond the scope of my project, I think.  I reserve the right for a long chew, or long bout of paralysis by analysis.

What I'm building is the kind of thing that would spit out DDL (i.e. the create statements for tables, indexes, views, foreign/primary/etc. constraints), the kind of stuff to create a database.  Not the kind of stuff for interacting with a database via SQL (select, insert, update, delete).

Using TiddlyWiki to access a database has been simmering in the far recesses of my mind, but that is a complicated affair, a whole bunch of technical skills I don't have at all.

However, if we are talking get TiddlyWiki to generate SQL statements that can be cut and pasted into other software that accesses a database, that could be a fun project, although likely one for a year or two down the road.  If TiddlyWiki can spit out DDL, it can equally spit out SQL.

So many possible projects, so few hours in a day.  Other projects I've been tossing around in my head: a TiddlyWiki Report builder (kind of like Crystal Reports?), and a TiddlyWiki Forms Designer.

[Hans Wobbe]

I certainly respect the right for a "long chew"... ... ...

Once you have a chance to swallow, feel free to email to kick around a few tentative first steps.  I will get a bit of an R&D budget allocated.

Cheers,

Hans