Follow

If you use a simple, text-based format for your data, your future self may be very grateful.

@praxeology

I generally agree.

But some open binary formats at remarkably stable, like my beloved SQLite, which is immensely more reliable to store tabular data that CSV.
(Don't use CSV, it's a trap).

@sebsauvage Right on. I use SQLite a lot too.

I'm just saying that if you have some data in a well formed text format (even annoying, inflexible CSV), you will never regret it in 5 or 10 years. Fancy binary formats, dependent on specific versions of specific libraries, you may be very sad.

@sebsauvage @praxeology As already mentioned, plain-text formats are inherently human-readable and therefore superior to binary formats regarding robustness. Also I propose the following (unproven) theorem:

For datasets smaller than 10k records (i.e. everything for personal use), the time difference between parsing a binary format and parsing a (well-defined) plain-text format into memory is negligible on current hardware.

@praxeology I moved my TODO and note systems to text files and it pays dividends year over year. No more lock-in, unlimited portability.

@annika @praxeology You like to use grep? No problem. ctags? Yup. Desktop search on Mac? Ditto. vespa.ai? Insane, unless you type a lot faster than what is considered physiologically possible, but also perfectly fine.

Sign in to participate in the conversation
post.lurk.org

Welcome to post.lurk.org, an instance for discussions around cultural freedom, experimental, new media art, net and computational culture, and things like that.