The case of the curious commit message | Hacker News Reader

The case of the curious commit message

chrisoldwood.blogspot.com

37 pointsingve14 comments

The old-school VCSes would force you to check out a file before you could start using it. I never used ClearCase, but we used to use VSS and Perforce. Both of these would lock the file on your machine, until you "checked it out." You could configure them to act as the case he describes acted, but we never bothered.

The need to "check out" a file, first, became rapidly infuriating. People would deal with it by simply leaving their files unlocked (using a shell script). "Checking out" a file would simply nudge the filesystem lock, as long as the server had not detected a change. In the case there was no change, the programmer could make changes, "check out" the file, just before checking it back in, and the server would get the change. If the server had detected a change, however (it did not use checksums, like Git. The server had a transaction database that it consulted), it would overwrite the file on the programmer's workstation, before "unlocking" it.

As you can imagine, this allowed me to vastly improve my vocabulary of swear words, in several languages.

I haven't used such systems, but I'm not surprised by the outcome. Telling a developer they can't edit a text file on their own computer is a system just begging to be worked around.

Git is wonderful. It took a cranky genius to make it.

BG (Before Git) was a dark time, indeed.

That's an excellent gotcha discovery.

The post says "On the plus side this got people discussing what a good commit message looked like". For git commit messages, and making them easy, it turns out a git commit template can help teammates ramp up:

https://github.com/joelparkerhenderson/git-commit-template

I have used both ClearCase and ClearQuest at a previous job. The former is surprisingly pleasant to use, although we had a wrapper program called GVP (Gather Validate Publish) [1] that handles commit messages and regression testing before publishing to "master". ClearQuest itself also has a very 1990's X11 GUI you can use that has a surprising amount of functionality.

The best feature of ClearCase is that with mvfs you can clone the repo in seconds since it's a virtual filesystem. A small database records your changes, and functions as a sort of overlay on top of the version you've "checked out".

ClearQuest was nice in that you can write your own SQL select queries to create bug lists. I don't remember if you could write DB-specific stuff (Oracle in our case) or if it had to be ODBC compliant. Unfortunately everything else about it kinda sucks.

[1] https://sourceforge.net/projects/releaseworks/

First thing that came to mind when I saw nt was “no text” shorthand (or no topic). Can’t quite pick where but suspect from IRC a long time ago.

I cant imagine a huge use case on IRC, but ive definitely seen nt used in old usenet posts

That was my first assumption as well. I'd seen it used (and used it myself) in usenet posts and email where the message was short enough to fit in the subject line, so you'd just append "[nt]" or something similar to end to note or confirm that there was nothing to read in the body.

I definitely used this on 90s-era forums.

Ah yes, forums, usenet and emails!

“nt” is, in fact, how you make a classic low effort shitpost. You make the thread title something provocative, and then the content is just nt.

At my (very small) college there was an email system where you could send emails to the entire campus. It pretty quickly turned into something that felt very similar to an online forum. It was conventional to use N/T at the end of subject lines for "No Text" when the subject line was all you needed to convey your message.

Email was checked with enough frequency that it was commonly used for what we use SMS / text-messaging for today. You didn't really expect people to be near their landline phone, but you could connect with most students on campus pretty easily/rapidly with the e-mail system.

This is an era before there was cell-phone service on campus, but cell phones were still very rare anyway.

This began as a story about the importance of well-worded commit messages. It ended as yet another testimony against pop-ups.

You know, I've clicked on so many things that just popped up while trying to do something else that I'd welcome some sort of global UI setting that made newly-presented clickable UI elements insensitive for 300ms or so.

Reply on news.ycombinator.com