XREF from XJCD

I apologise for this as I was looking at converting this to a mariadb db and all the frills and trimmings of a webpage but... I paid for this year long subscription to test this thing and haven't had a use for it yet, that it could handle properly.

I did nothing to this code, except upload it to claude fable5 and ask ->
"Can you convert this to a standalone modern javascript application or web based application?"
I haven't had AI do anything worth while but maybe it has this time.

So... yeah. I didn't even look at the new underlying code but it seems to work.

Unzip it and run it, the whole thing is embedded in a single webpage.

Have a go and report!

 

@cds1984 - can confirm, it works!! That's pretty awesome. Good use for AI.

I just have one thing on my wish list: It would be great if it could somehow include the parts fiche images.

 

I looked at the Derby database and shied away a couple of years ago. Didn't see any images of fiche at the time. I'll ask the bastard though haha. Not like I'm doing any actual work.
Let you know.

 

looks like there is no fiche imagery in there. I've asked it to add current information.
Bit of a stretch but who knows right!

 

Worth a shot!

 

hmmm okay.
Bit disappointing since I can't upload as the file is now larger (3.1mb) but...

Had a look and it looks the same roughly but yeah... no idea (not real trusting of this process where LLM/AI steals info that may or may not be correct)

Let me know via dm with an email address and I'll send it though via email if you want it.

I have no idea what is has done but here is the ... conversation, so to speak,

is it possible to add current information to this?

Weighed data sourcing constraints and devised multiple integration pathways

Good question — the honest answer is "yes, with caveats," because it depends on what kind of current information you mean.

The core limitation is that Yamaha doesn't publish an open parts API. Current fiche data, supersessions, prices, and availability live inside dealer systems and retail sites like Partzilla, Babbitt's, and Boats.net, and there's no legitimate feed I can bulk-import from. The 2006 snapshot in your Derby database is actually the last era when this data circulated freely as a standalone file — that's essentially why this app existed.

That said, there are three realistic ways I can extend the app, each solving a different version of "current":

1. Live retailer links — the easiest and most useful. Each part page gets buttons like "Check on Partzilla" / "Boats.net" that open a live search for that exact part number. You always see today's price, availability, and any supersession the retailer knows about, without me needing any data feed. This works because most Yamaha part numbers from this era are still orderable.
2. A data import/merge tool — a screen in the app where you paste or upload CSV rows (new models, parts, supersessions) and they merge into the database, with an option to export the updated file. Useful if you find newer fiche data yourself or want to add post-2006 bikes by hand.
3. Personal notes and overrides — annotate any part ("superseded to X per dealer, 2026", "NLA", "fits with modification"), saved in your browser and exportable, so your own current knowledge accumulates on top of the old data.

Which of these would be useful to you?



Q: What should I add to the app? A: All three

"
Searched the web

Partzilla has direct product URLs (/product/yamaha/PART) — even better than a search page. Let me check Boats.net:

Identified retailer patterns and architected implementation strategy

Both use clean /product/yamaha/PART deep links. Now let me implement all three features:

Verified completion and prepared documentation for release

All green. Let me update the README and ship it:

Ran a command, read a file

All three features are in — same single file, still fully offline for everything except the retailer links.

Live retailer links: every part page now has Partzilla, Boats.net, eBay, and Google buttons. I verified the URL formats: Partzilla and Boats.net both support direct deep links to the exact part number (/product/yamaha/90501-12083-00), so one click lands you on today's price and stock rather than a search page. Handy side effect: if a number was superseded after 2006, the retailer page usually tells you and substitutes the current one.

Import & merge: a new "My data" button in the header opens a page where you can paste rows (tab- or comma-separated, straight from a spreadsheet) for new models, part numbers, supersessions, or fitments. Rows are validated — duplicate models and parts get skipped with a line-by-line report, and fitments require the model to exist first. Merged data shows up everywhere in the app tagged "added by you," including in supersession chains and search.

Notes & overrides: on any part page you can set a status (Available / No longer available / Superseded), record a replacement number that becomes a clickable link, and keep free-form notes. Noted parts get a status badge, and all your notes are listed together on the My data page.

One design decision worth knowing: the factory data embedded in the file is never modified — your additions live in a separate layer the browser stores locally. That's why the My data page has Export my additions, which saves your entire layer as a small .json you can back up, move to another machine, or share with other owners, and Import additions file to load one. If your browser ever blocks local storage (some do for files opened directly), the app detects it, warns you, and the export button becomes your save mechanism.

I regression-tested everything end to end — imports, notes, the additions layer appearing in bike/part views, and all the original functionality still passing against the verified 2006 data.

 

Interesting. Would we need to access data from Boats.net though? Nobody has outboard motors on their bikes, I don't think.

 

Muwahaha... Are you sure about that?
Seriously though I have seen the boats.net listings when I've done part number searches.
At any rate DM me with the email and you can see what shite Claude AI produced. I'm not an advocate but it might be entertaining or useful.

 

I think boats.net and partzilla are the same thing

 

Well it seems to have other stuff like adding your own data.
Can't really see any external links.
kinda the same as the first version from that perspective. At least that one was small enough to upload.
I'll leave it at that.

 

Try this one. (zip file attached)
I threw it the file listing for 1/PARTS_DIAGRAMS and 2/TECHNICAL_PUBLICATIONS from the xjcd on my HDD and it added a library feature and hyperlinks on models etc.

If you save the file into the root directory of your xjcd data, on a computer it will point to the parts and service manuals also.

All the links are from my xjcd so it should be standard but you can change the location of the data also (per browser), from the library area.

I have to stop fiddling with this.
Would have taken me days to write this and months to debug the original I bet.
Being just a html page with javascript, you can jump in and alter anything with a decent text editor, whole database from the original xref is a single base64 string and the documents are preset arrays.
Madness.

 

Hey, nice work that you are attempting to do, and your instincts about LLM's is spot on, especially the word "steal", which is going to become a real sticky, thorny problem once the hype train ends and reality sets in.....and as usual, that will happen "later" (too late) rather than "sooner" (like, right now, when the problem might be able to be nipped in the bud).

Behold:

The part the AI industry does not want anyone thinking about:

https://nitter.poast.org/Ric_RTP/status/2072403984304984202#m

Palantir's CEO just exposed Sam Altman and Dario Amodei for robbing every Fortune 500 company. Within two minutes, Alex Karp took the entire frontier AI industry apart on national television.

His exact words:

"Every single enterprise in this country, these people are LIVID. They are paying for tokens that create no value. These people are stealing the weights and alpha of my business."

He literally said the entire frontier AI business model is intellectual property extraction dressed up as a subscription.

Then he also destroyed the pricing model with a single question that Silicon Valley still refuses to answer:

"If it was so valuable, let's say I can make you $1 billion tomorrow. Wouldn't I say I'll make you $1 billion and I want 30 percent? Why are they charging for tokens if it's so valuable?"

That question breaks the industry.

If OpenAI and Anthropic's models truly delivered the productivity gains the labs claim, they would take equity or a share of the profit they generate. They would not sell access by the million tokens.

Token pricing is itself the CONFESSION that the product cannot produce reliable value at scale. If it did, they would price for the value. But they price for the compute because that is what they are actually selling.

Karp went even further...

He called the entire arrangement "a wealth tax that does not help the poor. It just punishes."

American businesses are transferring the alpha of their operations --- meaning the workflows, the customer data, the strategy memos, the internal models that make them competitive --- directly into the training pipelines of a handful of Silicon Valley labs. Once those labs retrain, the customer's own edge becomes the next enterprise product sold back to their competitors.

And the part the AI industry does not want anyone thinking about:

Every enterprise running its confidential documents, its customer conversations, and its financial models through a frontier model is potentially teaching that model HOW to replace them.

The vendor collects the token fee AND the compounding intelligence about that customer's business. That is the mechanism. And that is why Karp used the word "stealing."

He claims this is why every executive he meets is furious in private and silent in public. Nobody wants to be the CEO who called out the labs and then discovered their next competitor was built on their own leaked workflows.

The entire AI industry has been priced for perfection on one assumption:

That frontier labs produce durable, defensible value that justifies infinite compute spend.

But Karp just told us that the customers do not believe that assumption anymore. They believe they are being taxed without benefit, watched without consent, and copied without recourse.

The moment enterprises stop believing, the whole valuation stack shakes.

https://video.twimg.com/ext_tw_video/2072403048115150849/pu/vid/avc1/1916x1080/tnyVGBCJ3Ll96WZI.mp4


---------------------------------------------- MORE ----------------------------------------------

Claude AI For Dummies:

Letter to an accounting firm regarding this article (which is a good article):


Hi Joe, I thought this was a pretty good article on how to use AI more effectively:

https://ruben.substack.com/p/claude-for-dummies


Len
-------------------------------

(Joe responds, "thanks for the info!)

-------------------------------

Hi Joe, hopefully that will help with any of the AI services you use.

You probably already realize this, but a word of caution:

• “File work. Claude’s desktop app can see your local folders (from your computer) and work with your files directly. ChatGPT can’t (yet).


Just be aware that this lets a private company browse thru your computer (you’ve authorized it to do so) and that means that any info that you’d prefer to keep private/secure (which may be many in the accounting biz!) is no longer so, as you have authorized Anthropic (or whomever) to read thru them under their agreement, and even worse…..although not explicitly noted……it quite possibly / probably is storing that info on their servers (so that the AI doesn’t have to go re-research with every query from you)….this is of course much more time- and storage-efficient for them, but……..your data in your machine(s) are now “backed-up” to an outside source.

This leads to a couple of possible scenarios which are worrisome:

a) Anthropic (Claude) gets hacked (intentionally or not) by someone else, and all your data now goes to them, too…….

b) Your data and info from your computer can possibly now be used (and for a moment, even forget about it being due to a hack) by someone else when doing their own, different query to the AI. For example:

- someone else enters a query: “tell me everything you know about Joe Blow of Blow Accounting Services”.

Now that is a very broad query, but if the data from your computers, which Claude has saved due to your past interactions with it, then it can be used by Claude to provide a response to someone else.

And as the article suggests, that person can now drill down with more specific queries, such as:

- “can you explore whether there are any inconsistencies, fraud, or possibly illegal activities that are occurring with Joe Blow / Blow Accounting Services or his clients?”

or:

- “tell me where Joe Blow / Blow Accounting Services or his clients might be financially (or otherwise) vulnerable to pressure (i.e. blackmail or otherwise)?”


Now, the sellers of AI services will tell that you that their models are programmed as to not engage in or respond to such (morally or legally restricted) queries, but…….with just a modicum amount of “clever” questioning you can “trick” AI into bypassing those (supposed) safeguards and provide almost any and everything that you might want know about anyone or anything….since that information and data is already “out there”, (or “in there”….in their servers, that is) and all it takes is for an AI model to put 2+2 together via clever “queries” to spill the beans about it.

After all, that what AI is all about……gathering every bit of data in existence (that goes by the cutesy term of “scraping” data from every source possible that it can hoover up----legally or not) and then creating the ability to put “answers” back to a user who asks it to show relationships between any and multiple datapoints that exist.


You see where I’m going with this…although Anthropic (etc.) may claim that they don’t save or store this data, it would be ridiculous from a business profitability / computer resource usage for them NOT to do so. It would be like re-inventing the wheel on each query! That’s the purpose, after all, for all these gargantuan “data centers” …. to store and process data. Since AI models require “training”…...which means data capture and storage for future response…and storage of data on a mass scale is both cheap and basically what any “AI” is all about...…then you can wager a good amount with confidence that is exactly what is going on….even if they deny it is (because, otherwise, the whole idea of AI is fallacious).

Upon reflection, AI models seem to be a self-selected, self-authorized version of full-spectrum technological “spying” which the marketing departments wrap in a pretty package in order to get you to ALLOW yourself to be spied upon.


Remember, all sorts of people said that the benefit of Bitcoin (or any cryptocurrency) is that it provides “anonymity” in regards to financial transactions…..a complete lie, it does exactly the opposite….it records every financial transaction of anyone who participates in it from the very first initial transaction with it until “death do us part”….and beyond.

So upon reflection, cryptocurrencies and their core “blockchain” technology seem to be a self-selected, self-authorized version of full-spectrum technological “financial spying” which the marketing departments wrap in a pretty package in order to get you to ALLOW yourself to be spied upon.



Cloud computing (and back-ups) portray themselves as “secure”, when they are anything but:

You might find these articles interesting, this guy started and ran one of the earliest ISP’s in the country.

https://market-ticker.org/akcs-www?singlepost=3414912

And in case you’re interested in “Part 1” of the above article……..

https://market-ticker.org/akcs-www?post=231578

So upon reflection, “cloud computing” and their core technology seem to be a self-selected, self-authorized version of full-spectrum technological “spying” which the marketing departments wrap in a pretty package in order to get you to ALLOW yourself to be spied upon.


The joys of modern life……….I suppose you may have seen the 1980’s movie “Wall Street” with Michael Douglas and Charlie Sheen……although what most people recall from the movie is the Gorden Gecko character’s “speech” saying “Greed is good”, I believe the most important part was this:




---------------------------------------------- ONE MORE...FROM THIS WEBSITE ----------------------------------------------

The “mosaic” strategy, where missile commanders move and attack at their discretion, has an extra brilliance:

AI will not be able to model it, and ALWAYS give false positives, because as I noted off-topic a couple of times, one of AI’s weaknesses is if you feed a scribble to it, and ask what letter it is, it will give an answer with 99% chance confidence it is correct, because “not a letter” is beyond its capability.

Right now I read the IDF is saying things are getting better, evidenced by there was an 8-hour period with no attacks on Israel, and there was no barrage greater than 100 missiles/drones.

They are looking for patterns where there are none. Their AIs will make predictions of patterns it sees in the randomness, unable to understand what randomness truly is.


Brhatweed:
I tend to shy away from the AI, most of this is based on pure speculation and strings of data put together by lazy teenagers who think life is a game.


---------------------------------------------- CLOSING THOUGHTS ----------------------------------------------

“The underlying purpose of AI is to allow wealth to access skill while removing from the skilled the ability to access wealth.”

 

How many times have you spoken to someone and just off the cuff they will have the answer/explanation to any subject/question.
Generally known as a bullshit artist.

That's what LLM is at the end of the day. If you believed everything that is published to the web you would have a contradictory mindset but the LLM model scrapes the web and vomits the information back to the user as gospel, with the caveat that it makes mistakes.

So my original apology stands for sure.

About a month ago I was tasked to implement a function to be written by a third party programmer based on Claude AI. Completely spurious function to allow voice searching and a voice replies.
After the guy quit because he couldn't achieve what I wanted, I went through the code and realised that every single line of code and documentation was written by the LLM.
The actual guy couldn't code at all and made this clear at the end.

Obviously this was upsetting.

At any rate. I wanted to see if I was becoming obsolete, thus the subscription purchase.

I tossed it some of my custom code from the early 2000s and said modernise it.

Total garbage and functionally different output.
Can't think if it's too old.

Asked it to fix a simple piece of code. Returned code with old functionality and deprecated syntax.
Can't think of it's too new.

It doesn't think.

At any rate it is a tool. I won't be renewing I can tell you that.