[Welcome to Sensei's Library!]

StartingPoints
ReferenceSection
About


Referenced by
WikiOrphans
LinksInDiagrams

 

Links in diagrams - History
    Keywords: SL description

Moved from Opening Systematic Classification.


Blake: Apologies if this is a dead concept, but I don't check Meta much... I just read the joseki library bit down at 15 Mar. What if there were a special sort of pagetype, 'joseki,' which would isolate the joseki pages from the rest of SL? they could be reached by something like senseis.xmp.net/?joseki=4_4=3_6 , where the numbers are the spaces played (4_4 3_6 would refer to the hoshi with a keima approach, etc.). A system like this would be necessary, I think, to avoid bogging SL down in pages like "JosekiHoshiKeimaWhiteSanSan?." I'm not sure how the SL wiki code works--I've not had a look at it--but this doesn't seem particularly Evil, and probably wouldn't require editing of existing pages. Thoughts?

Morten Blake - there has been some attempt at this already. The 'Joseki' keyword is used quite consistently on Joseki pages, as has the '3443etcetera' form for the title. There is some inconsistency regarding whether 3 is spelt 3 or three, and there was some discussion about whether '446346' was more understandable than 44lowapproachikkentobi, but no consensus was reached - see OpeningSystematicClassification.

Blake - Ah, I see. My approach was more aimed at a technical solution than a 'readable' solution. I was thinking about having some sort of imagemap or, preferably, a simple board diagram which one could click on to follow the joseki--similar to kogo's, but with the important difference that it allows discussion. I must say, I'm not a fan of the current system. But I'm just a lowly 18k, what do I know? ;)

Morten - There is sort of a beginning attempt at a hierarchical breakdown on the Joseki page itself. I cannot see us adding imagemaps to the wikiengine in the foreseeable future, but it may be useful to add illustrating diagrams on this page. Then again, having said that and thinking about it, imagemaps should be quite easy to add. To make the potential benefits clearer, Can you explain to me what you would like to achieve, why you cannot do that with today's version of the wikiengine and how imagemaps would help?

Arno: Blake, the imagemap idea is nice, but hard to fit into the current wiki structure. I think what really is needed is a "search position" utility. I will attack this problem sometime this year. I think of allowing "stored searches", so that you can in fact define a diagram (e.g. a joseki) and search for all pages where similar positions appear. Once this feature is implemented I think one can add nice overview and index pages for all kinds of positions.

Blake: Maybe 'imagemap' is the wrong term. I was thinking of an extension to the current diagram format which allows variations to be marked, and played out on other pages. Currently, clicking on a diagram allows you to get a .sgf file, which seems like an under-use of it. Perhaps clicking on the diagram title can get you a sgf file, while clicking on an actual space can get you a variant. Variants could be defined in the diagram system. A neat solution to the joseki problem, and it seems as though it could have uses elsewhere.

Actually, it could be setup so that clicking on any diagram automatically makes a page for a variant. the wiki could name it something like "PageName?_DiagramTitle?_Variant", and mark the variant with a circle on the diagram... though this could get VERY unwieldy with sub-variants and so forth.

mAsterdam or maybe something like this (This is an early draft. There is a modified example below):

(There is no diagram version of the first example below. It kills the page for low-res viewers. Blake removed it)

 $$B Reference
 $$  ------------------
 $$ | . . . . . . . . .
 $$ | . . . . . . . . .
 $$ | . . . 1 . . . . .
 $$ | . . . , . . . . .
 $$ | . . [a|#L1] [b|#L2] [c|Gomoku Outside Keima#Dia1]. . . . .
 $$ | . . . . . . . . .

[Diagram]
Referenced 1

 $$[L1]B Referenced 1
 $$  ------------------
 $$ | . . . . . . . . .
 $$ | . . . . . . . . .
 $$ | . . . 1 . . . . .
 $$ | . . . , . . . . .
 $$ | . . 2 3 . . . . .
 $$ | . . . . . . . . .

[Diagram]
Referenced 2

 $$[L2]B Referenced 2
 $$  ------------------
 $$ | . . . . . . . . .
 $$ | . . . . . . . . .
 $$ | . . . 1 . . . . .
 $$ | . . . , . . . . .
 $$ | . . 3 2 a . . . .
 $$ | . . . . . . . . .

c in dia 1 would refer to a diagram on another page


Blake: I don't know; the links being -inside- the diagram doesn't do it for me. The diagram structure needs to stay as simple as possible. Perhaps, after the diagram's structure is done, there could be a line like this:

 $$ V:x:y:z:L

Where x and y are (obviously) the x, y coordinates (relative to the diagram) and # is the 'variant number' to be placed there. If it's not given a number, it could be marked with a circle. Thoughts? L could be an optional link.

mAsterdam: Hmm... It creates two references for every link: one to the diagram coordinate, one to where the browser should jump when somebody clicks at that point. I fear that this extra level of indirection will be error-prone. But it looks better. This is important, too.


Discussion moved to Bob and JF on readability.


Morten I see the two issues of imagemaps and josekinames as quite distinct.

Imagemaps can reasonably easy be implemented as suggested by Blake, although I wouldn't advocate the use of the 'z' parameter, as any board markup is already given in the diagram. That leaves a notation along the lines of

 $$ V:xy:HREF     example: $$ V:C3:UpperManchurianFuseki.

In some cases the link may be to another page, or they can be to a bookmark on the same page. We can always argue about whether it should be 'V' for variation, 'I' for 'Imagemap' or 'L' for link :-). In addition to being used for Joseki, this could also be used for problems, OngoingGames etc.

'Joseki Names' as used in pagetitles is another issue. As mentioned above, we must be reasonable. I do not think that having a unique identifier for a 69 move joseki should be a driver for defining the system we use.
The current use of 443664etc is systematic and logical, but I agree that it is not very appealing to look at. However, the only people who will always see links to the actual pagename (instead of a [pseudonym|443664etc]) are those who access the pages through recentchanges... All other users will see whatever pseudonymn is deemed suitable.
Obviously, once the page is displayed, the pagetitle is awkward, but this can be alleviated by consistently providing a diagram at the top of the page, making the title itself less important. (Already the case on many if not most Joseki pages). If this is combined with an easy way to click your way through variations (be it on different pages for major lines or on the same page for smaller ones), I think we will have achieved a lot?

Transposed positions are easiest dealt with by suggesting a consistent approach, e.g. 'Black plays first in the corner, white always approaches from the right (along the top)'

DougRidgway I like the idea of links in figures and figure captions. One way to do it would be to first enable links in figure captions, so that

[Diagram]
After the low approach, a?, b?, ... or tenuki? comes next.

works in the expected way. Then, intersections in the diagram referred to by name in the caption could be linked to the same place as the caption. The only disadvantage is that you have to mention the figure links in the caption, but this seems like good policy anyway.

Blake: I don't like having to mention the links in the caption :p Perhaps a letter or number (or circle, or square...) which is linked should be colored differently. Jago uses a small green cross to designate a variation; having, say, a green 'a' or a green circle seems logical. The only problem I see is that it would effectively double the number of images required for diagrams... but SL is hardly image-heavy as it is :) Imagine if you have a diagram that has 10 variations--not impossible in a joseki tree, or a problem set. Mentioning the letters in the caption would get very cumbersome, very quickly. Links in captions -are- a good idea, though.

DougRidgway: You're probably right about not requiring mention of links in the captions -- don't make the technology enforce particular style on the authors. Perhaps silent links could be added with link lines after the title line, like this:

 $$L [a|ThreeFourLowApproachOneSpaceLowPincerOnePointJump?]
 $$L [b|ThreeFourLowApproachOneSpaceLowPincerOutsideShoulderHit?]

I still think links should only be from named points, not raw coordinates. Two reasons: easy to author, and robust against edits in the figure which can alter coordinates.

Arno: this last proposal looks like a clean way to implement links. It does not interfere with the current board markup and the title line. I'm still unsure, whether links in diagrams is a good idea or not. Feedback from other deshis?

mAsterdam: It won't come as a surprise to anyone that I think it is a very good idea.

Right now the text is interwoven through hyperlinks. The boards (the diagrams) however are separate, secondary entities. The diagrams act as a prison in keeping the moves and boardpositions isolated from the rest of the material. The possibility to link the intersections to text and other diagrams would make the stones and intersections on the board an integral part of the web. This is very appropriate, especially at SL, a site dedicated to Go. The moves deserve to be integrated as a worthy part of the go-language.

So it's all about the ability to link from relevant places on this here diagram to another diagram, whereever it may be. Doug's proposal looks like easy editing.

details:

 $$L [a|ThreeFourLowApproachOneSpaceLowPincerOnePointJump?#Continuation 2]
 $$L [2|ThreeFourLowApproachOneSpaceLowPincerOutsideShoulderHit?]

Where "Continuation 2" would be the title of a diagram on ThreeFourLowApproachOneSpaceLowPincerOnePointJump?

2 would be linked, too. (Necessary? hmm I don't know, but I can imagine where it would help so why block it).

In this proposal only relevant (marked or played with a number) intersections can link. This is good, I think.

A visual clue to the reader that the spot is linked (this is important I think) could be underline.

The consequences make the example (testcase? usecase?) look like this. Even the 'L' looked redundant so I left it out. This discussion is about adding links to diagrams. Looking at the first example one can see that this is now exactly what happens: There are some links added to the first diagram.


[Diagram]
Reference

(links underscored)

 $$B Reference
 $$  ------------------
 $$ | . . . . . . . . .
 $$ | . . . . . . . . .
 $$ | . . . 1 . . . . .
 $$ | . . . , . . . . .
 $$ | . . a b c . . . .
 $$ | . . . . . . . . .
 $$ [a|#Referenced 1]
 $$ [b|#Referenced 2]
 $$ [c|Gomoku Outside Keima#Dia1]
 $$ [1|Gomoku]

[Diagram]
Referenced 1

 $$B Referenced 1
 $$  ------------------
 $$ | . . . . . . . . .
 $$ | . . . . . . . . .
 $$ | . . . 1 . . . . .
 $$ | . . . , . . . . .
 $$ | . . 2 3 . . . . .
 $$ | . . . . . . . . .

[Diagram]
Referenced 2

 $$B Referenced 2
 $$  ------------------
 $$ | . . . . . . . . .
 $$ | . . . . . . . . .
 $$ | . . . 1 . . . . .
 $$ | . . . , . . . . .
 $$ | . . 3 2 a . . . .
 $$ | . . . . . . . . .

c in dia 1 would refer to a diagram on another page




This is a copy of the living page "Links in diagrams - History" at Sensei's Library.
(OC) 2004 the Authors, published under the OpenContent License V1.0.