Earth Notes: On Website Technicals (2019-11)

Updated 2022-10-23 14:19 GMT.
By Damon Hart-Davis.
Tech updates: new Fairphone 3, MIDI data feed, GSC PageSpeed Insights, intensity log live, h3 tweak.
tools
From page speeds to style tweaks, and a new live datafeed (long overdue).

2019-11-24: H3 Style Tweak

I have been trying to resolve a minor style issue for a while. How should I capitalise my headings H1, H2 ... H6 especially as I am now typically using H3 for FAQ questions?

(I basically never use H5 and H6, and only rarely use H4.)

Typically, for headlines, I like to Capitalise All Significant Words. That makes sense for the single H1 that I have on each page for the title and headline.

It also makes sense for H2, which I use for the major sectional headings in most pages.

Capitalising all significant words looks a bit strange for questions in FAQs, especially when (for example) Google scoops them up into rich search results.

Also, until today, I have styled H1, H2 and H3 to be sans-serif.

From today, just H1 and H2 are sans-serif and should be the only all-significant-word capitalisation.

I feel better with that off my chest.

(I also created some additional log filtering today which shows that about half my pages, including most of these Site Technicals, are essentially never read. That accords with Google Search Console reporting them not in G's index.)

2019-11-18: Intensity Log Live

I have just made the GB electricity grid computation be logged live under data/FUELINST/log/live/.

This will allow retrospective more detailed calculations of local (16WW) carbon footprint from power flows ie generation, imports, exports and consumption.

Note that the logging rate will not be constant, eg will drop when available energy (ie the off-grid battery) is low. Logging will stop while fresh data is not available from Elexon.

A new dataset has been created (2019-11-23) too, to help make it discoverable.

2019-11-04: Moderate AMP Page Speed?

GSC (Google Search Console) has introduced an 'experimental' Speed sub-section of Enhancements today, it seems.

Happily it rates all the desktop (www.) pages that it is prepared to talk about as 'fast'.

However it rates all of what it calls my 'valid' mobile pages (just amp., not m.) as 'moderate', apparently in part because of the time to load the core AMP script!

The GSC-reported aggregate FCP (First Contentful Paint) value is far higher than PageSpeed Insights run directly on any of the GSC-mentioned 'moderate' pages (1.1s vs 0.7s).

I think that the ad load does not help either, so I'm trying cutting back ads on non-desktop pages generally. For example, not inserting a footer ad, which would likely be poorer value-per-byte than other ad slots.

I may drop ads from a larger set of less-popular non-desktop pages, to make them lighter and thus more likely to appear in search. In any case, if sufficiently far down the tail their revenue loss is likely to be insignificant.

I could in future give in and use AutoAds to more-dynamically adjust ad load. Though not being able to adjust desktop and non-desktop separately is not good.

2019-11-08: ... and today GSC is showing all listed (123) mobile (AMP) URLs as 'fast', with an FCP (First Contentful Paint) of 1.0s. (I pressed the 'verify fix' button a couple of days ago. Maybe the ad diet was enough.)

2019-11-03: MIDI as Dataset Feed!

Slightly whimsically, I have just been prompted to publish a MIDI (music) file as an alternate form of one of my datasets!

The MIDI file will be extended along with all the other versions as I add new data points.

Since I have all the scripts in place to do it, and it's quick, only a little makefile magic was needed to have everything update when a new data point is added.

I don't imagine that many datasets are published with an encodingFormat of audio/midi! (Though I see that there are many datasets that are collections of MIDI files.) Maybe that will cause a search-engineer a little debugging fun...

2019-11-01: Fairphone 3 Arrived!

Partly because Brexit did not happen last night, my new Fairphone 3 turned up today. (I had fears of it being entangled in whatever customs excitements would have cropped up if we'd crashed out of the EU.)

So after trying to stop Google listening in on everything by forbidding the "Google" app from accessing the mic, the next thing I did was find the screen size for responsive layouts, natch.

(I note that it is not possible to disable the voice assistant, constantly relaying your words to Google HQ, without signing in to a Google account, which I don't want to do for privacy and security reasons. Catch-22. If that is within the letter or spirit of GDPR then I am a pimple on a Belgian's behind...)

360x720 CSS pixels portrait, 720x360 landscape. (1080x2160 ie 2,332,800 real pixels, pixel ratio of 3, colour depth 24 bits.)

(My FP1 has 540x960 ie 518,400 pixels, pixel ratio 1, colour depth 32 bits.)

Thus on the correct side of my 640 CSS pixel-width first responsive breakpoint when the FP3 (or FP1) is in landscape.

Getting the bumper and back off and the battery out was easy. I was able to copy over contacts from my FP1 via microSD card, ie sneakernet. Unfortunately I need a smaller (nano?) SIM for the FP3 than in my FP1.

The battery arrived ~50% charged. Charging from my MacBook Air via the USB (A-to-C) lead I bought with the FP3 is reported as 'slow' on the FP3, but seems an OK speed to me.

I also discovered that the FP3 plays .mid (MIDI) files directly, and does 2G and 5G WiFi: woot!

2019-11-06: I continue to test pages on the FP1/Opera, via WiFi.

~1062 words.