My job described perfectly by Dilbert

[cjolley]

This is the best description of the daily workings my job I've ever seen.

[Claymonkey]

Lol!

[Brian Ellis]

Been there, done that

One project I did took 4 years of argy-bargy in the early 90s. In the end, I produced a magnificent product working under MS-DOS. It worked as perfectly as any software will work and was thoroughly idiot-proofed against what I thought was anything that my client could dream up. However, I delivered the final version and submitted my invoice (large). Three days later, the client phoned me, said that it looked good but they now wanted six more modifications, one of which was to change the OS from MS-DOS to the brand-new Windows 95. I politely explained that this was not possible as it was programmed in a version of Forth which was greatly modified for scientific applications and that there was no way I could transport it across platforms. From past experience with his changing his mind every 5 minutes, I said that it would mean a) I would have to learn how to use C/C++ for Windows, probably the Borland version, b) write the code for .h routines for at least a couple of dozen complex maths equations, including polynomial curve fitting with 4 different algorithms c) put the lot together. I told him that it would probably take a year but I would undertake it only if he agreed that it would have a fixed specification, according to the current working model, no other modifications and that he would agree to a charge of about $¼M. Silence at other end for ½ minute and he came back, with a croaky voice, "but, no, it would be part of the present contract, no extra money". He then withheld payment of a 6-figure debt to me. I started to get the legal eagles working on it. My client then sold his enterprise to a German company, concealing the fact that they had the debt to me (and to a few other people in similar positions, it transpired later). I claimed from the Germans and they were VERY surprised to hear I had 4 sealed contracts, would I please fax them copies (totalling about 70 pages)? Two weeks later, they sought a meeting with me and their lawyers. They were very apologetic, and we settled amicably out of court for about 80% of the debt plus future royalties plus legal costs incurred to that date, all of which left me in profit. And I didn't have to do a Win version! (In fact, I've never seriously tackled Windows programming, except a few minor things in VB).

[Tjalfe]

could you not just have made a pretty icon for it, and made it launch the dos program from windows?

[TransformX]

could you not just have made a pretty icon for it, and made it launch the dos program from windows?

Exactly what I had in mind

[Brian Ellis]

Nope (I tried that). Firstly, the Forth refused to work under DOS 7 (it was programmed under MS-DOS 6.0) Then the client was talking about having multiple graphics windows open in a different way to how it was done under DOS. Also, I looked at the possibility of NT. One of the big problems is that a single test could take up to 28 days. I had MS-DOS tamed for that with very careful programming with modules, so that I could free RAM at the end of each modular routine. My initial tests with WIN95 could not achieve the kind of reliability we needed, short of having each module in its own reserved RAM. I calculated this would require something like 512 Mb RAM: this would have cost a fortune in those days, even if the computer accepted it (this was pre-Pentium and we were working with 486/7 processors). If I remember correctly, we were limited to 4096 kbytes!!!!! It was only a couple of years earlier that Bill Gates said that 640 kb was enough for anyone!

[TransformX]

Nope (I tried that). Firstly, the Forth refused to work under DOS 7 (it was programmed under MS-DOS 6.0) Then the client was talking about having multiple graphics windows open in a different way to how it was done under DOS. Also, I looked at the possibility of NT. One of the big problems is that a single test could take up to 28 days. I had MS-DOS tamed for that with very careful programming with modules, so that I could free RAM at the end of each modular routine. My initial tests with WIN95 could not achieve the kind of reliability we needed, short of having each module in its own reserved RAM. I calculated this would require something like 512 Mb RAM: this would have cost a fortune in those days, even if the computer accepted it (this was pre-Pentium and we were working with 486/7 processors). If I remember correctly, we were limited to 4096 kbytes!!!!! It was only a couple of years earlier that Bill Gates said that 640 kb was enough for anyone!

I suspect 1/2 a gig of ram would still cost less than one year of you rewriting it..

[cjolley]

I suspect 1/2 a gig of ram would still cost less than one year of you rewriting it..

That was their problem, not his. He had already delivered.

When my wife was getting her masters in creative writing one of her professors said something wise:
A writer never really finishes a story, the writer just eventually abandons it.

Programming is sort of the same way.
Especially in the eye of the consumer.

You just have to tell them that they have what they said they wanted, anything new is a new project.

Of course what they say they want is almost never what they really want.

I tell our new programmers to keep in mind that all projects have have three sets of business rules:

1, What they say they want it to do.
2, What they should want it to do.
3, And what they really want it to do.

These three are NEVER the same. :sigh:

And since they will continuously orbit mentally amongst the three, they will never see the project as complete, because their fantasy about it will never stop changing.

[KvHagedorn]

To create the ideal program for a business (if such exists) would require the programmer having worked in that business for many years.. but then he wouldn't be a programmer.

[cjolley]

While this isn't a business, I've worked here for seven years and this stuff still comes up from time to time.
Things I have never heard of, even though they have always done it (perhaps on paper) and have always wanted it automated, yet never told the programmer who works 30 feet away. Some of these types of things have taken less that 10 minutes to code up.

Then there are the once a year things that seem to change rules each year for no dicernable reason.
User: "Chuck the yearly "X" doesn't work right! It leaves interest out of the total!"
Me: "Thats the way it worked last year."
User: "It can't have! It would be illegal!"
Me: "And for the preceding six years."
User: "It's got to be fixed ASAP."
Me: "Ok, how should it work?"

--A day later
User: "Chuck it's wrong again! It puts interest in the total!"
Me: "I know, it was tricky, but I got it in there without rewriting the whole thing from scratch."
User: "It can't be. Interest can't be in the total. It would be illegal!"
Me out loud: "No problem."
Me to myself: "Ok, commenting it out again, like every year."


Sigh. again....

[Technoid]

Oh man, I'm happy I only have to discurage people wanting to use a DVI>VGA adapter with a standard vga cable and then a VGA>DVI adapter at the monitor end
(since he had DVI conectors at both end I eventually after a loooong discusion convinced him to just buy a DVI cable )

[Sasq]

brian, cjolly, we should get jw and have a chat in irc one day

see above comic, see dan and jw's time sheets, see 300+ hour months on our time sheets.

[gt40]

I'm soooo glad I abandoned programming 15 yrs ago
and not for the reasons most folks think - I was extremely good at it - just couldn't stand sitting by the screen for 10+ hrs a day, and then being told I needed to rewrite because they changed their minds

Btw, how many of you have programmed in APL?

[schmosef]

I have a bunch of Dilbert's taped to my office door. The one from this thread would be on the door too if my colour printer wasn't currently buried under a half ton of clutter. I'll try to remember to dig it out later so I can print it.

[Brian Ellis]

I suspect 1/2 a gig of ram would still cost less than one year of you rewriting it..

Maybe, but not when it was used in hundreds of places, world-wide. The computer was part of a complex scientific instrument, where it was used to number-crunch the results of up to 100 sets per minute of 6 measurements over 28 days and to present the results in real time in a way that someone looking at the screen could understand at a glance. When the test was finished, the salient data on HDD, with a redundant back-up HDD, the full analysis of the data was done and it took the computer several hours to do this with a colour print-out of about 800 pages. This was split in ten parts, with a pause to change both the black and colour cartridges in an Epson Pro-XL printer at every other pause and the paper at every pause!!! Our initial trials used a colour matrix printer but the graphics were not good enough and the printhead seized up too often because of the heat from such intensive use.