Undocumented features

[redbird]

That's programmer/hacker snark, as in "it's not a bug, it's an undocumented feature" when the code doesn't do what it's supposed to.

Only sometimes there really are undocumented features: deliberately created by the programmers, tested, doing what they're supposed to, and part of the running software, but deliberately not documented.

In this case, I've run across one that is deliberately undocumented, but will be explained to anyone who writes to tech support and says "I want to do this thing". As far as I can tell, this has the effect of adding to the workload of the support staff, and rewarding people who either don't read the documentation or assume that it's incomplete, while depriving people who figure that if it's not in the documentation, they can't do it.

I can't see how this is a win for anyone.

The feature in question isn't one I have any particular interest in--it controls how much of your LiveJournal [1] is syndicated via RSS. But (a) this strikes me as a bad idea, and (b) I wonder what else they have tucked away that I might want to use.


[1] more precisely, the publicly visible posts therein

Comments

[adrian-turtle.livejournal.com]

I can't speak for programmers, but features in more tangible products are sometimes left undocumented because the developers didn't finish developing or testing them until after the deadline for the regular documentation. "Should we publish a special bulletin to tell all the users about this feature?" "It's a nice little feature, but only a handful of users are likely to care about it, so it's not worth the trouble and expense of getting the word out to everyone. We can just tell the people who ask."

Sometimes a new feature is valuable because it's a partial fix of the old system. Marketing people are oddly reluctant to publicize something in the direction of, "New! Improved! Breaks less often than our old product!" There are customers out there who are content with the old product, who might not notice the breakage problem unless you told them about it.

I don't know how well the reasoning works, but it's reasoning, it's not just a thought-gap.

undocumented features

[pir-anha.livejournal.com]

I can't speak for programmers

i can, for some of them. :) and you got it in one. minus the "nice little feature" thing, really -- if it's not totally tested before the "cut" deadline, it won't be announced in the large graphics package with which i work.

[cakmpls.livejournal.com]

I really like you. Have I ever told you that? I do.

You are a ittle bit like me, and a whole lot not like me, and women who can do techy stuff are hot.

(I'm monogamous and straight, but I know hot when I see it.)

[redbird]

Thank you.

[lisajulie.livejournal.com]

There's also the issue (not particularly in LiveJournal as I best can see) of having features that work, but that the folk behind the scenes don't want to support.

After all, having something cool and useful is neat. But answering eleventy-dozen emails/phone calls/etc. about "how to do this/why to do this/should I do this? can cause a developer to reconsider making features public.

[almeda.livejournal.com]

Another, possibly, that I use every day but can't find in the docs anywhere:

Recent Comments (http://www.livejournal.com/tools/recent_comments.bml). Only works if logged in. Staggeringly useful to me in particularly because LJ stopped emailing me any comments about two years ago, and nothing we can try can make it start doing it again.

[akicif.livejournal.com]

Hmmm. Would it tie in with this (http://www.livejournal.com/users/marnanel/645512.html), by any chance?

[redbird]

Yes. I saw rho's post, and then the support_interim post about synlevel.

[liv]

On second thoughts, I probably shouldn't rant about this stuff publicly. But the general message remains: I agree entirely with this post and I'm feeling pretty frustrated myself.

[redbird]

Understood. At moments like this, though, I'm glad I get emailed replies.

I haven't done much support lately for other reasons, as you know--health, then work, and that I'm spending my random evening online time IM'ing with adrian_turtle instead of on the support board--but things like this do not encourage me to continue/return.

[mneme]

One reason (aside from "won't break anything, but not ready for prime time") that things like this get left out of the documentation is that including too many features can sometimes make the documentation much harder to read and follow, whereas giving a much more limited example (frex) can produce more maintainable documentation. Possibly not the case here, and not something I ever choose to do as a developer (but something that's occasionally forced on me by eternal circumstance), but there you go.