kitty_keys VS epub-create

Ah yes! i forgot about column(1), although the different implementations have different behaviour, which might or might not be an issue for the OP.

But although i've done quite a bit of programming and sysadmin stuff since then, it's only more recently that i've really started exploring the more nitty-gritty sides of shell scripting, as a personal challenge, albeit with a focus more on portability than on nifty tricks - cf. e.g. my kitty_keys function, or my epub-create script. Which shellcheck(1) had some useful complaints about, but at least one thing i disagreed with, based on my personal preferences derived from my decades of programming. :-)

It's in the repo i linked to upthread; in particular, please refer to this file at this commit, with the four instances of cat ${SPINE} | replaced with <${SPINE}[a].

In my own case, although i'm not active on any microblogging platform, i am a FOSS dev and, er, 'documenter', and i recently read up a bit on the EPUB standards to put together a guide and shell script for creating a minimal EPUB.

Last year i encountered a problem caused by LVM's build system relying on a bashism. And i recently release a simple shell script for creating EPUBs, which tries to avoid using non-POSIX functionality so that it's more likely to be usable on different systems/platforms.

i can reproduce this on my system, using my epub-create script i released just the other day. :-)

In the context of the Ruby snippet you shared, the indentation doesn't matter for what the code is doing, but in other contexts, the indentation *can( matter. For example, i recently wrote a shell script to create a minimal EPUB, which uses heredocs to create XML and XHTML files. If Emacs indented the 'headers' and 'footers' according to their position in the code structure, their indentation in the resulting files would be misleading (even though this wouldn't affect the validity of the markup).