Hacker News Reader

Comment by enscr | Hacker News Reader

When I'm arranging my photographs in folders, I prefer yyyy-mm-dd. That way it stays in chronological order.

However for casual reference, when year is not needed, I find dd-mm more intuitive. I guess it has something to do with reading from left to right... at a glance, it's just easier to comprehend.

I've always found mm-dd a bit odd to use.

View on HN

It's an international standard: https://en.wikipedia.org/wiki/ISO_8601

There are two ways to send dates across the wire as text between different systems:

1) ISO 8601

2) A world of hurt

Source: experience. Not using the standard invites things to break, but only on a some machines, and only on or after the 13th of the month. Of course the data for the first few days of the month will be present, just wrong.

And always - always - always in UTC.

And sadly there are still things that insist on using the second option. RFC 2822 used in MIME (historical oddity because it pre-dates ISO 8601), UNIX timestamps as either seconds or milliseconds (at work we have one instance where numbers > 32 bit are serialized as a hex string in JSON, which includes such a millisecond-UNIX-epoch timestamp), etc.

Slightly related insanity: Using localtime (i.e. not UTC) for log timestamps. And then having fun once a year dissecting the one hour that has twice as many messages.

It's annoys me that the time format with colons means I cannot use it for file names. What do you use instead?

Some other innocuous separator character such as underscore, or none, just fixed lengths for the numeric fields. But always, yyyy mm dd order.

Underscores work well, which results in "2013-12-16 10_22_57Z". I can still somfortably read it, at least.

Hyphen-minus would lead to too easy confusion with the date part. Fixed-width fields can be hard to read.

I just use HHMM at the end (or HHMMSS if you need that precision).

Ex: 2013-12-16_1741.txt and the like.

> When I'm arranging my photographs in folders, I prefer yyyy-mm-dd. That way it stays in chronological order.

It is kind of interesting to think about that from a usability perspective, because the "stays in chronological order" refers to how most file system browsers default to sorting things by name. If it was a concious decision to sort by name, then the description would be something more like, "when I am sorting my folders by name, I'd rather sort by date, so I put the date in the name."

Folders do tend to have Date Created metadata, so in theory you could "sort folders by created date" rather than by name, but where that data is stored and how that data changes as the files are moved around is pretty opaque, so it makes lots of sense to not rely on it.

Date created might not match the metadata date for the content. For example, you might download a bank statement for november, in december.

When I'm copying/modifying/mirroring folders across platforms, the date in metadata can get messed up. However, a name like "2013-02-22 Trip to Tahoe" is what I find most useful. Each folder has a mnemonic & I can search/browse my pictures chronologically when I'm feeling nostalgic :)

Sometimes the natural-seeming title of a document includes a date, but it isn't the creation date (simple example, scanning a letter or invoice). Would you suggest frobbing the metadata?

The advantage of using yyyy-mm-dd, beside the fact it is a standard, is that it is unambiguous. When I read a date like 12-11-2013, I don't know if we're talking about the 12th of november as we French correctly write it, or about the 11th of december as those bloody arrogant US American incorrectly do.

But 2013-11-12 is non-ambiguous since (I guess) nobody uses a yyyy-dd-mm format.

Reply on news.ycombinator.com