A metaphor stretched past the breaking point, which might be funny unless you're arachnaphobic

[azurelunatic]

I think the reason I am pretty decent as a person who locates bugs in programs is because I like to understand how things work. (It is only relatively recently that I have both attained confidence at ad-hoc speaking and begun to know so many people that I cannot maintain working models of all of them in my head, so I can rehearse conversations ahead of time so I won't appear stupid or weird by encountering a situation that I have no words for.) If a program's response confounds the way I expect the thing to work, I try to find out why, and find out what's supposed to happen. This results in me poking the thing from many directions, figuring out how to make it happen again, reading the documentation on what's supposed to happen, and by the time I go to the people who are supposed to make things Just Work, I have a lot of things to say.

This, of course, makes me incredibly popular at work.

Today's contribution to my general "popularity" with the AAAA WHY DID YOU DO THAT IT'S FULL OF SPIDERS OR SOMETHING group involves a lot of very boring detail work. I am going back through every goddamn study my department has logged since like 2008 and confirming the following things:

* There is a link in the database to the wiki page that the study is about. (The converse, a link from the wiki to the database, cannot happen because THE DEVELOPERS HAVE BEEN SMOKING SPIDERS, and you cannot link to an internal database page, despite the database being web-based. I have filed an enhancement request for to be able to link. I expect it will be denied on the grounds that THIS IS TOO FANCY FOR THAT AND WHY WOULD YOU WANT IT BY THE WAY WE HAVE CONSUMED CAFFEINE WITH THESE SPIDERS AND WE CANNOT WEB.) I neglected to do this on the first year or two. I shall have to go back and fix.
* There is a start date associated with the project. (This is a requirement enforced by the database. You'd think all of them would. However, one of the projects has the year "0000". Yes, an import of data from a previous instance of the database was involved. No, they did not have me go through it all in Excel first. They should have. No, I cannot correct it. Not for lack of trying, but because it will not physically let me save it. I have filed a ticket.)
* The number of participants logged to the project in the database should jive with what it says on the wiki. Sometimes there is no record of "participated", just "considered". This is rough when the numbers don't add up.
* The dates of participation should be generally harmonious with the start date of the study, unless it's indicated that it's a long-ranging study. Unfortunately, this data is very, very, very dirty. As in, stuff from 2009 sometimes has a date that looks like the date that something was uploaded en masse to the new database, namely sometime last December. (This, coincidentally, may have been from the terrified upload where I noticed that there'd been some data loss of stuff I knew I'd entered.)
* Every study is associated with a piece of software. (More or less.) Generally speaking, every participant has experience with the piece of software that the study is about. (Also more or less.) More saliently, this experience is supposed to be logged in the database. Guess how many people this is true for. The current round of swearing and bug-filing is because I know my Overlady is going to ask me, probably next week, for all the people who we know who have had experience with Product X. ("Product X" sounds like something that could melt your face off. Let me assure you that the good folks at Virtual Hammer do not create tools that will melt off faces. At least, not customer-facing tools.) Therefore, I am going through by brute force and making this happen. Yes, I am aware that a few joins and an update query would probably make my life a lot simpler. Am I going to get the access to do a few joins and an update query? Fuck me running, but I'd even love a read-only query access to this thing. Am I going to get that? As if. But I am "enjoying" the tour through the data, and it's providing me with plenty of chewy software bugs. AND SOFTWARE SPIDERS.

And then there was that one guy who was entered twice in one very old study without multiple participation, and now not only can I not update the second record, but when I try to edit it, it pulls up the contents of the first record. I rebooted my machine to see if there was some sort of weird cacheing thing, but no, it really is doing that. I'd be inclined to ignore it, since it's really obvious and also from 2009, but I have the eight-legged crawling fear that it's a symptom of a really stupid programming choice that someone made, and that if I don't report it now as a problem, that it'll turn out to be venomous when it bites me in the ass later.

The great (and perhaps slightly sad) part of all this is that if I hadn't noticed that it was after 10pm by reason of the lights going out and starting to get hungry again, I might have stayed later, because I was enjoying tromping around with the data-machete.

Comments

[vass]

I have been reading too much Thor fic lately, so now your tag is making me imagine you as a blue-haired Mjolnir.

Also, I'm going to steal THE DEVELOPERS HAVE BEEN SMOKING SPIDERS expression and use it every time I have to log into certain government websites over here.

[azurelunatic]

I imagine the spiders are none too thrilled.

They CANNOT WEB.

[azurelunatic]

(Regarding experience, sometimes studies are done with first-time users.)

[silveradept]

I sincerely hope Virtual Hammer appreciates you and this work, and without spiders involved.

[azurelunatic]

All my work on this thing is sort of limited impact, because it's a Very Special Tool. The results of improvement to it is increased efficiency from me, and keeping my researchers happy.

The work on the other program is appreciated such that when it went from being in-house to being out-house (heh, heh, heh) the designer explicitly mentioned that I knew where the product's public Bugzilla was, and invited me to continue my "guerrilla QA".

I'm awesome.

[silveradept]

That was never in doubt. The degree of awesome, however, is probably greater than what Virtual Hammer is used to getting, so I really do hope they fund it in Ryder wallets top reward you properly for these things.

[jamoche]

This results in me poking the thing from many directions

This is what sets you above my testing team. They poke things very hard, sure, but from only one direction at a time.

[azurelunatic]

And possibly they never sit down and think "What possible other directions could I poke this from?" when they commence work, or find that something breaks when they poke it?

[jamoche]

Given that they find edge cases like "switching between A and B as fast as possible on a very slow machine might crash about 1 in 20 times" but don't find "after switching between A and B at least twice, B doesn't behave correctly" (and guess which one the users are going to hit) I wonder if they think at all.

[wibbble]

I am 100% certain that Virtual Hammer, Inc are not paying you enough, regardless of how much they're paying you now. This DW entry should be linked from your CV.

Also, I even went and looked at their product list and couldn't figure out (other than maybe their Mac desktop app?) which one you thing has a face-melting name.

[azurelunatic]

The placeholder name "Product X" is the one that sounds face-melty. Something mysterious that might or might not cause unexpected effects to the unprepared.

A raise would be very nice if they felt like giving me one! I'm trying to get into High Availability mode, because I'm not currently physically capable of working all the hours available to me every week, and I'm sure that more things to do would emerge if I ran out.

[wibbble]

Ah, I get it now. :o)

[jamoche]

If Az was finding bugs in that app, trust me, I'd be a much happier dev.

[azurelunatic]

Full-time QA honestly sounds like the sort of thing that I would find fascinating for several weeks, and as soon as I'd started feeling comfortable doing it, I'd start going completely out of my mind with stress and boredom (depending on what kind of week it was).

Due to the tabs I have open, I am now picturing attempting to train hens to QA. The resulting mental image is hilarious.

[jamoche]

QA is doing the same thing over and over again and expecting different results, and you know what people say about that :)

[azurelunatic]

Indeed.

[azurelunatic]

What I Would Do If I Had Read-Only Query Access, by Azz, Aged 33 1/4

First I would glare at the table of studies. Next I would select every user who has participated in a study. For each study+user, I would retrieve the user's first and last name and the program associated with the study. Then I would sort by the user's name, whichever one of them the user table is displayed in order of when I go look at it.

Then I would manually go in, user by user, and make sure that all of them have the programs listed that the query says they should have.

I'm still getting it done "the fun way", though.

[wibbble]

If you can find someone who really understands 'devops' as a philosophy they might be able to act as an advocate for getting you this kind of access.

[aedifica]

O'Reilly is (was?) having one of their ebook sales again, and I picked up one on Bad Data just because it sounded fun. Not having bad data, of course, the book.

[azurelunatic]

Ooo, that does sound fun.