Humans maintaining autogenerated code? (11)

1 Name: #!/usr/bin/anonymous : 2006-11-13 03:39 ID:5hsZcm6+

I understand that in corporate environments it is common to have programs autogenerate code, and then have humans (in theory) maintain that code. Is this true? Does it work well, or is it a disaster?

As a Unix-style hobbyist programmer, I see that as a bad idea. There's nothing wrong with having programs write code, but I think it should be considered an intermediate step in the building process, not part of your source code. And it should just be automatically updated on a build after you change the code-writing program's input. That seems a lot more maintainable. Am I failing to see something here, or do corporate programmers just use a bad approach?

2 Name: #!/usr/bin/anonymous : 2006-11-13 04:01 ID:Heaven

> do corporate programmers just use a bad approach?

yes.

3 Name: #!/usr/bin/anonymous : 2006-11-14 09:29 ID:Heaven

>>1
it's not so bad if you're paid by the hour...

4 Name: #!/usr/bin/anonymous : 2006-11-15 00:04 ID:Xvkz8THZ

>>1
I doubt very much that it's true.

Perhaps you're thinking of stuff like Visual Studio and Access's application wizards, which give you a barebones to start writing your program in? They all generate stuff that you'd never need to change.

5 Name: #!/usr/bin/anonymous : 2006-11-16 13:27 ID:Heaven

>>4

He is definitely not thinking about that. I am not familiar enough to explain exactly what kind of tools are used, but there are many toolsets you will not encounter outside the corporate programming world.

6 Name: #!/usr/bin/anonymous : 2007-01-12 02:38 ID:pKKtEIWu

>>1

There are some systems that will take a software design (UML) and write code for it. Said code has certain benefits - consistent (if dull) naming conventions, type-safety, syntax-bug free, etc. For very boring business-logic type applications, it can be a shortcut. On the whole, I tend to view it as a bear on a unicycle - an entertaining trick with very little redeeming value. We may come to a point 15 or 20 years from now where "programmers" can debug an abstraction like UML, but for now you always have to go into the code. I wouldn't describe it as "common", however - at least not in the places I have worked.

7 Name: #!/usr/bin/anonymous : 2007-01-17 20:52 ID:p7D6xlJF

UML is a big fail because of all the bloat and redundancy (20 different types to name the same thing) it brings in, when I learned some people actually want to create UML VM i was like O_o ... No human ever will be possible to maintain this pile of crap once it runs.

But alot of programmers use other cool code generators such as wizards or visual designers to create forms (or another stuff that has no impact on core of the program anyway).

Also think how the life would suck without (f)lex!

8 Name: #!/usr/bin/anonymous : 2007-01-18 21:00 ID:oV5IPTZd

>>6

Anything that can be translated into code must contain all the same rules as the code, which means it is just as hard to write.

http://www-inst.eecs.berkeley.edu/~maratb/readings/NoSilverBullet.html

9 Name: #!/usr/bin/anonymous : 2007-01-19 17:13 ID:5hsZcm6+

>>7
But nobody modifies the C code generated by lex. You just modify the .l file, and the computer updates the .c file for you as part of the building process.

10 Name: #!/usr/bin/anonymous : 2007-01-22 07:52 ID:+I0N9/Ll

Computer Scientists becoming keeps for machines. I guess that's the way it was supposed to turn out.

11 Name: dmpk2k!hinhT6kz2E : 2007-01-22 19:09 ID:Heaven

System administrators are the machines' keepers. Software engineers write the software that system administrators curse. And computer scientists design the languages and algorithms that the software engineers ignore.

This thread has been closed. You cannot post in this thread any longer.