OS_GBPB9 in practice

I semi-regularly see people using a verbose description as a filename.

photo-of-john-smith-standing-next-to-the-avon-river-on-23-april-2006-holding-a-microphone-while-giving-a-presentation-to-lincoln-high-school-students-on-the-importance-of-cleaning-and-drying-the-bottoms-of-boats-to-prevent-the-spread-of-didymo-in-our-waterways-7b.jpg

think it’s good to not break at a space, but good to break at a hyphen, comma, apostrophe, @ and many other things.

Is that by intention, or did somebody get their if clause logic back to front by accident?

That being said, it’s an easy trap to call into – Netsurf is quite happy to wrap around to a new line on the apostrophe between the “can” and the “t”…

I semi-regularly see people using a verbose description as a filename.

A problem addressed by my Pic_Index program ( link ) where a description can be attached to an image in a form easily searchable by keyword!

Just sayin’.

Is that by intention, or did somebody get their if clause logic back to front by accident?

Logic? This MS we’re talking about.
Since it’s been like that in every version I can recall this century, there’s probably options:

1. Deliberate, wilful stupidity.
2. Lack of common sense or general working grey cells

The problem is that it is often required to set a buffer size, it is only less of a problem as there is plenty of memory available nowadays, tough luck for older machines that would otherwise benefit.
The prospect of UTF-8 in filenames means you would have to allow 4 times what you would think was a generous amount, so I’m ending up with 1Kb per name 16 entries at a time = 16Kb
I think you would get away with 1k for 16 entries until you looked at foreign machines multimedia files or www filenames.

UTF-8 in filenames means you would have to allow 4 times what you would think was a generous amount

This is not unreasonable, though in lots of places the limits work the other way – Module names for example are a maximum of 39 bytes long, which as I’ve often remarked is only just enough for this one ‘character’:

One of the things I was very pleased with in the Twemoji project was a way of turning these huge Unicode sequences into valid sprite names… and back again. So the Unicode sequence here, and therefore its full textual name, is encoded in the twelve bytes of the sprite name:

Anyway yes, UTF-8 means Make No Assumptions.

Though having said that, MetaSprite means one can easily embed any amount of data inside a sprite, so that was a brief victory.

Just looked at the Unicode 15 emoji list, and its characters to bytes ratio is 3.1428. So perhaps PI is a good multiplier!

Sigh.

I note with sadness that all versions of FontManager get GBPB,10 wrong – erroneously believing that R4<0 means ‘finished’.

Sadness? Really? You were expecting Acorn to follow their own rules?

;)

A quick hack for you to try if you’re so inclined.
A little module to hijack GBPB 10 (etc) to return with R4 bit 31 set (and strip it on entry so the underlying FS won’t know anything has been fiddled with).
See how much stuff then breaks because “less than zero means done”.

You were expecting Acorn to follow their own rules?

Based on 35 years’ evidence, I’m not expecting them to have even heard of them.

A quick hack

My HeebyGeeby module does this randomly.

sadness

Aaaaaand now I realise that FontManager thinks “Z123456789” is a metrics file.

Peer review? Pier revue more like.