Gah! Stupid fucking programmers!

The BBQ Pit
41

[QUOTE=Orbifold]
I’m siding with Otto here. Switching to wildcard-based searching strikes me as a classic example of forcing the user to conform to the program instead of the other way around. Not to mention a violation of the KISS principle. Offloading work onto the user is not a good design principle.
[/QUOTE]

Maybe they replaced an in-house developed search module with one that comes off-the-shelf, has more functionality, and costs much less to implement and maintain. If so, that’s a great design principle. It saves money so that people can keep their jobs.

I’d also like to add that if the software were always 100% static and deterministic, then you could be replaced by an IVR in a matter of months. Humans do jobs that computers can’t do. Stop complaining that your employer is playing to your human strengths and reinforcing your job security.

42

[QUOTE=Otto]
Whatever dick. Enjoy your little life.
[/QUOTE]
Well, you’re the idiot who started an entire thread to complain about having to add an asterisk where you didn’t have to before, which will occasionally cost you a second but save you many more seconds equally often. Cock.

43

Everyone should read Cosmic Relief’s first post. 100 Doper Bucks says it’s all about the speed. Speed speed speed speed speed. Wildcard searches are far more inefficient than exact searches, and the table of customer names is probably one of the biggest in the system. If your previous searches were all returning partial matches, then they were all using wildcards (as others have said) - now you have to deliberately choose to do so, and a lot of people won’t. This will free up resources on the system. Maybe you’ll notice a speed improvement yourself - maybe not, maybe the programmers are using those extra cycles to implement some new groovy feature that you personally don’t even use. But I’m dead certain they wouldn’t be futzing round with a piece of working functionality unless there was some reason - even if you don’t see it - that they had to.

44

[QUOTE=Digital Stimulus]
Perhaps. As I mentioned above, there might be an “optimize for the common case” consideration at play here. But – and I may have a blind spot here – it’s not at all clear to me that using an asterisk to signify “any match” is a difficult concept.
[/QUOTE]

I don’t know if the problem is so much that it’s ‘difficult’ as that it’s ‘unusual’. I haven’t seen an asterisk in a search query since the days when you could exit Windows into DOS. And as someone else pointed out, people tend to be really stupid when it comes to computers (it doesn’t help that the computers are even dumber and only do exactly what they’re told). I can easily imagine someone who doesn’t understand wildcards simply refusing to use them, because it’s not what they’re used to.

45

[QUOTE=Cosmic Relief]
Hey, I was a developer on that system. The reason you lost access is that the “quick” way used a grotesquely inefficient database query that would bring other systems to a screeching halt for several hundred users. But as long as you showed good judgment, we let you use it. Then one day your team hired a dingdong who kept hitting it every 10 seconds just to see it refresh. Then we had to restrict access and from that point on we weren’t allowed to develop any more “don’t push this button” buttons unless they were restricted to a secret ninja team who understood why it shouldn’t be abused. Sucks, but that’s how it is.
[/QUOTE]

I suspect we are talking about two completely different systems (cough DSL cough). I’m talking about being able to check the network card and connections directly. No doubt somebody was messing with it improperly, but that happens with several thousand tech support guys. Once we lost that, the whole thing turned into a huge slog as no one could get the real information they needed. More to the point, cutting access totally is really bad “solution” because it makes the situation worse.

But A Certain Southern Phone Company wasn’t too keen on actually getting things done. I loved the fact that we would be fired for not following directions exactly or not getting calls done (on average) withing the required time limits. Of course, the official directions included several useless or near-useless steps, which we were supposed to do in a set (nonsensical) order, and was invaraible impossible to succeed with.

46

[QUOTE=smiling bandit]
I suspect we are talking about two completely different systems (cough DSL cough). I’m talking about being able to check the network card and connections directly. No doubt somebody was messing with it improperly, but that happens with several thousand tech support guys. Once we lost that, the whole thing turned into a huge slog as no one could get the real information they needed. More to the point, cutting access totally is really bad “solution” because it makes the situation worse.
[/QUOTE]

Oh, I see, it’s a different system than what I thought. In this case there is an element management system (EMS) that talks directly to the cards, and a network management system that commands the EMS. There are a couple of different problems in play.

In some cases the EMS is poorly designed and becomes crashy when too many people are logged into it.

The more pervasive case is when you get heroes who go in and start manually manipulating cards, ports, and connections in the EMS. The problem is, the NMS is the system of record, and if it doesn’t have an accurate representation of what’s been done in the EMS, then Bad Things Start Happening. NMS talks to Every System, as in billing, physical plant, inventory, tech support. When it gets bound up, it affects the entire world. As in, hundreds or thousands of customers getting randomly disconnected by routine orders to connect, disconnect, or move services, service appointments start getting missed, people get overbilled or underbilled, connection commitment dates start getting missed. Then it takes a Very Large Crisis Team to sort out the problem caused by this one hero who thought he was hot stuff trying a secret ninja move to sort out one customer.

The overarching problem here is that DSL, at least in its early implementations, was way too complex. The original assumption was “We’ll just hire smart people and make sure they only do the right things”. But the support burden grew too fast, and the complexity grew too fast. It wasn’t possible to add more smart people that quickly, and the technology changed so fast that it wasn’t possible to keep everyone as smart as they needed to be. Things have simplified somewhat, but the supporting systems still show all the fingerprints of that initial overengineering.

Not that I know anything about your actual situation, mind you, I’m just taking a random guess based on your description.

47

Since we are pitting programmers, I pit lazy or short sighted programmers who can’t deal with two word surnames. My surname is of the form “van Surname”. This is translated into “Firstname Vansurname”, “Firstname Van”, etc. My driving license reads “VAN-SURNAME” thanks to an inability to include spaces.

One of the only institutions to get this right in the US is immigration, which makes sense I guess.

48

[QUOTE=Cosmic Relief]
The overarching problem here is that DSL, at least in its early implementations, was way too complex. The original assumption was “We’ll just hire smart people and make sure they only do the right things”. But the support burden grew too fast, and the complexity grew too fast. It wasn’t possible to add more smart people that quickly, and the technology changed so fast that it wasn’t possible to keep everyone as smart as they needed to be. Things have simplified somewhat, but the supporting systems still show all the fingerprints of that initial overengineering.
[/QUOTE]

Oh god, I hear that!

Troubleshooting DSL is particularly challenging because there are a lot of points of break. Cable is usually pretty stable because the technology is well-known; the major fail-points are in just a couple small areas and more easily fixed. But DSL is a more complex beast with a lot of interactions between software, hardware, and modem.

At one point we could check to see if the network cards on our end were responding. If not, we knew the problem was there. We could also use them to ping the house modem. If not, the problem was there. This was a hugely useful tool, since our major first step was to reboot the modem. But if that failed, they took away our ability to gather further information.

This was probably because their off-site tech support included several different companies. Some of them had more or less reasonable hiring practices (us), although some bad apples got in anyway. But the level of indifference, ignorance, and poor training at the others was legendary. Mostly, they did not even try to resolve problems, because they got judged (and therefore paid) not by fixed systems, but by calls resolved in the “official”, prescribed manner. Any further troubleshooting was a waste to these people.

The irony was that A Certain Southern Phone Company went well out of their way to make things tougher. The ancient client record system was grotesquely slow because it simply hadn’t been designed for the level of use they had. It would have been reasonably easy to upgrade, but for some reason they couldn’t spare the money for a new sever (dunno why). Of course, everything required us to use that and we wound up having it crash twice or thrice a week.

49

[QUOTE=Orbifold]
I’m siding with Otto here. Switching to wildcard-based searching strikes me as a classic example of forcing the user to conform to the program instead of the other way around. Not to mention a violation of the KISS principle. Offloading work onto the user is not a good design principle.
[/QUOTE]

I see the exact opposite. I’m quite a believer in KISS, (and in the Law of Least Astonishment) and if I type Robert Smith into a search, I want to get all the Robert Smiths, and not Roberta Smith or Roberto Smith or Robert Smithson or Robertson Smithfield. I’d also say that sorting through all the false positives is a lot more work than typing in a “*” when you actually want the extra matches.

Our command line name and phone number search works the way the old db does, and it is a pain - but I can pipe the results through grep or more if I want. That’s more work, and requires more knowledge, than just allowing wildcards.

50

[QUOTE=Driver8]
Since we are pitting programmers, I pit lazy or short sighted programmers who can’t deal with two word surnames. My surname is of the form “van Surname”. This is translated into “Firstname Vansurname”, “Firstname Van”, etc. My driving license reads “VAN-SURNAME” thanks to an inability to include spaces.

One of the only institutions to get this right in the US is immigration, which makes sense I guess.
[/QUOTE]

Now that would be a worthy pitting. But the real people to pit aren’t the programmers, but the clowns who designed the system and wrote the requirements assuming that your type of name does not exist.

51

Children, old people, and idiot users of computer systems don’t like change.

End of discussion.

It doesn’t matter if it’s faster, or more efficient, or better. It’s different, and to the vast herd of sheep who use these things, that’s all that matters.

I am just as bad, but I also understand that usually there is a reason for the changes.

52

[QUOTE=Voyager]
I see the exact opposite. I’m quite a believer in KISS, (and in the Law of Least Astonishment) and if I type Robert Smith into a search, I want to get all the Robert Smiths, and not Roberta Smith or Roberto Smith or Robert Smithson or Robertson Smithfield. I’d also say that sorting through all the false positives is a lot more work than typing in a “*” when you actually want the extra matches.
[/QUOTE]

I agree. The new search system seems a hell of a lot better than the old one to me. I suppose the other way they could have changed it is to keep the default wildcard option on, but to force exact matches with quotation marks around terms you want exactly matched.

For example ROB SMITH would return ROB A. SMITH, ROBERT SMITH, ROBERTA MARIE SMITHWICK, etc, but “ROB” SMITH would only return ROB SMITH, ROB A SMITH, and ROB THELONIUS SMITHERS. ROB “SMITH” would give ROB A. SMITH, and ROBERT SMITH and, finally, “ROB SMITH” would only return ROB SMITH.

Either way works, but a forced wildcard search would drive me batshit insane.

edit: Oh, and one more option: “ROB” “SMITH” would return ROB SMITH, ROB A. SMITH, and ROB MICHAEL SMITH.

53

[QUOTE=Miller]
Moreover, IIRC, Otto’s position is a temp position, which I suspect means that there’s a fair degree of turn-over.
[/QUOTE]

I’ve been a permanent employee for almost two years now, but there are a lot of temps and a lot of turnover.

And searching SMITH ROB*T still returns Roberta and Robertonia.

54

[QUOTE=Otto]

And searching SMITH ROB*T still returns Roberta and Robertonia.
[/QUOTE]

Ah, that’s like the old Commodore 64 wildcard that ignores everything after the asterisk. I wonder whether this is more common or less common wildcard usage.

55

[QUOTE=pulykamell]
Ah, that’s like the old Commodore 64 wildcard that ignores everything after the asterisk. I wonder whether this is more common or less common wildcard usage.
[/QUOTE]

Less common. It certainly doesn’t follow standard UNIX regexp rules.
Now it is okay to pit the programmer because he/she must have been too lazy to do it right.

56

[QUOTE=Otto]
And searching SMITH ROB*T still returns Roberta and Robertonia.
[/QUOTE]
Well, that’s downright bizarre. (Does it return middle names on them too?)

57

[QUOTE=Voyager]
Less common. It certainly doesn’t follow standard UNIX regexp rules.
[/QUOTE]

Nitpick: this looks less like regex and more like globbing.

Eh, that’s all I got.

58

[QUOTE=Tristan]
Children, old people, and idiot users of computer systems don’t like change.
[/QUOTE]

I give you this smiley :slight_smile: because I have worked with many of the people you reference above.

But on a serious note, the problem is really documentation. And, this is an area where programmers seem to really suck balls. Ok, you changed how wildcards worked. So fucking tell us, so we don’t have to sit there and figure it out while a frustrated customer is looming over us. Ok, you changed what information we can access. Howzabout you let us know that so we don’t tell a customer we can access that information when we really cannot?

My current job is helping a professor teach a computer course - we are entirely driven by the software. The folks that make this software implemented some significant changes a couple of weekends ago. Overall, every change is a very definite improvement - the problem is that they didn’t fucking bother to TELL US what had changed. As a result, I was very nearly maimed by a horde of disgruntled undergrads. Its not really the students’ fault - they were suddenly confronted by software which did not work the same as when they last accessed it. The fault lies squarely with the programmers for not bothering to let anyone know.

I fully appreciate that you have improved my software. I would be even more appreciative to have you realize that I am not a fucking psychic.

59

[QUOTE=Subway Prophet]
Nitpick: this looks less like regex and more like globbing.

Eh, that’s all I got.
[/QUOTE]

I glob all the time - I do a lot of Perl. Also, ksh uses globbing for filename completion. (You can use standard regexp expressions also.) But it doesn’t make a lot of sense to use that in the OP’s context.

I figure if you are defining a search, you might as well do it the way the rest of the world does.