Irreal

I’ve written before about the ed editor and the faithful who still use it today. A tweet by Mike Zamansky lead me to this Julia Evans post on batch editing files with ed. Evans has a typical editing problem: she wants to duplicate each line in several files that contain a certain string but also change the string to something else on the new line. See her post for the details.

She chose to do this by writing an ed script to make the change and then call ed with the script (presumably) as part of a larger shell script that does this for each file. My first thought was that this made sense even today because ed scripts have always been used in just this way and it’s still an easy and elegant solution. But then I got to wondering how I’d perform that task.

My inclination was to try some sort of Emacs trick, of course. Evans is a Vim user so she doesn’t have the awesome power of Emacs at her fingertips and can be forgiven for falling back to ed. Since we haven’t had an Emacs puzzle for a while, here’s a challenge: figure out how you’d use Emacs to get this done.

Zamansky, being less inclined to rush in, offers the following excellent Sed solution:

Motivated me to play with sed – cat f | sed -e "/baz/P" -e 's/baz/elephant/'

— Mike Zamansky (@zamansky) May 12, 2018

It’s probably hard to beat that solution even with Emacs but see what you can do.

I feel almost like a troll writing about this. The controversy as to whether there should be one space or two after a period is surely one of the most pointless of holy wars. Yet holy war it is. For those of you can came in late—that is, who grew up using computers instead of typewriters—the issue is whether typists should put one or two spaces after the period ending a sentence. The two spaces made sense for typewriters with their fixed-width fonts as it made the text slightly easier to read. When computers came along and everyone started using proportional fonts, the second space was no longer needed. At least that’s what the typography experts said.

Nonetheless, the debate raged on. Having learned to type on actual typewriters, I was for many years in the two space camp, if only out of habit. Around 2011, I changed to using a single space, adjusted my Emacs to recognize that as a sentence end (as related here), and have been happily conserving spaces ever since.

One of the reasons I say the argument is mostly pointless is that when documents are typeset, the composition software is probably going to ignore that extra space anyway. That’s something Avi Selk found out when he tried to insert an extra space in the headline to this Washington Post article on the controversy: it broke the formatting. And, sorry two-spacers, once he got it working it looked weird. Follow the link and see for yourself.

Both Selk’s article and a a similar one in Ars Technica by Sean Gallagher report the results of a study published in Attention, Perception, & Psychophysics that found two spaces slightly increased reading speed in some subjects.

Leaving aside the fact that the majority of psychological studies are not reproducible and therefore suspect, there’s plenty to criticize about this study on its own terms. Most glaringly, to me, is that the study tested its subjects using the fixed-width font Courier New 14. The researchers justified this on the grounds that the font was standard for this type of study but that misses the point: Of course the extra space helps with typewriter-like fonts. Everyone agrees about that; the question is does it make sense with proportional fonts. In any event, the modest increase in reading speed (≈ 10 words/minute) was seen only by those who used two spaces.

So the TL;DR is that we don’t really know more than we did before and the debate will continue unabated. The Washington Post article appears to favor the two-spacer camp while the Ars Technical article seems partial to the one-spacers. Go read both articles; there’s something for everyone.