Tools: add crude calculator for transaction size

[harding]

This adds a crude Javascript-based transaction size calculator to /en/tools/calc-size/. No links are added to the tool from other parts of the site and I'm not sure we should add any unless the tool is improved.

However, I've found it good enough for my own use and I think it might help y'all when you're burdened with verifying my transaction size math in newsletter and workbook content. Future improvements can let it deal with a wider range of multisig variations, differing types of input and output scripts, non-keypath P2TR, uncompressed pubkeys, low-R signatures, LN onchain scripts, and other common transaction types.

That said, if you don't think this is a good fit for the Optech website, let me know and I can easily host it somewhere else.

[jnewbery]

I think it's fine to add this. I also agree that we shouldn't link to it until it's a bit more polished.

One thing that I'd like to see added is documentation on how the values are calculated.

[harding]

One thing that I'd like to see added is documentation on how the values are calculated.

I don't understand this. Is this more code documentation (I thought that was ok), more description about what fields are included in a transaction in the HTML output, or something else?

[jnewbery]

I don't understand this. Is this more code documentation (I thought that was ok), more description about what fields are included in a transaction in the HTML output, or something else?

More description in the html about how the values are calculated.

[bitschmidty]

tACK

Cool tool! Left a couple suggestion.

[harding]

Forced push, hopefully addressing all comments (including providing more information about how the values are calculated). Thanks for the reviews, @jnewbery and @bitschmidty!

[bitschmidty]

@harding I think the legend is really informational. Im also using the table calcs in a presentation right now!

Resolved previous comments and added a few new ones.

[0xB10C]

I like this.

[0xB10C]

Both HTML form inputs could do with a min=0 attribute.

[0xB10C]

That would be cool! I might do a follow up PR that adds this.

[harding]

That'd be great! Also having someone work on this who actually knows JS would probably be very beneficial. :-)

[harding]

Force pushed some edits:

• Hopefully resolved @bitschmidty's editing feedback
• added min="1" to input and output count form fields (@0xB10C suggested min=0, but since consensus rules require all transactions contain at least one input and one output, I decided to use a minimum of 1)
• Separated the size calculation logic function from the HTML presentation function
• Added some crude semi-automated testing to verify the form is generating accurate results. I only tested the default 1-in, 2-out cases; linking each of them (except P2WSH) to an actual transaction on the block chain. This revealed two minor errors, which I corrected:
  • P2SH 2-of-3 was off by 2 vbytes because its scriptSig field size uses a 3-byte compactSize uint instead of the 1-byte uint I expected
  • P2WSH 2-of-3 was off by 0.25 vbytes because I forgot to include the size of the witnessScript field in the calculation (which would've been obvious if I read my own documentation on the page). Additionally, using a quick scan of blocks 600,000 to 601,000, I was unable to find a P2WSH 2-of-3 transaction with 1-in, 2-outs that used 72 byte signatures for both of its sigs, so I don't have a txid reference for that. However, I'm sure it's correct based on a mixed 71/72 byte signature transaction I checked.
• I also disabled breadcrumbs for this page because it wants to link to an /en/tools/ page that doesn't exist. Obviously we can add that page later if we want to publicize this

[harding]

P2SH 2-of-3 was off by 2 vbytes because its scriptSig field size uses a 3-byte compactSize uint instead of the 1-byte uint I expected

Oh, perhaps worth a dorky note, I think that if at least one signature both signatures for P2SH 2-of-3 use low-R (and all pubkeys are compressed), then this can be a 1-byte compactSize uint. Obviously, anyone who cares about space savings should be using P2WSH instead, but it might be worth a footnote in the scaling workbook or somewhere else that using low-R sigs can save you up to 4 vbytes here instead of just the expected 2 vbytes. 🤓

[bitschmidty]

tACK 6d0712c (Chrome, Safari, Firefox on macOS)

[0xB10C]

tAck 6d0712c

• tested in Firefox, Chrome and Chromium
• looked over the size definitions
• checked for dead links (automatically)

[harding]

Pushed a single line edit to make the P2WSH 2-of-3 witness example more accurate and less confusing.

[jnewbery]

Looks great. Thanks Dave!