Thread: learning c from old k&r code

i found these files and started to study and update /rewrite them to learn from it.
V7/usr/src/cmd/yes.c

i have however a question. i noticed there is no license .no comments, and besides from the old code it also uses many goto statements in some of the sources.

i been under the impressing that only when writing on a kernel would ever need a goto in code and i also think from having read some other sources for the same programs from solaris and from gnu core-utils that these old codes seems to be not so well written and somehow very simple and i wonder if it is actual old production code.

if anyone has similar links to older code or code like this i can "steal" and learn from i would be very happy. especially if the sources are available used for exact that purpose.

thanks.

what would the purpose of rewriting some already functional code?

I ask myself that every day when I come to work.

But why - modern code doesn't HAVE to fit in 64KB, or run quickly on a 4MHz/3-100 clocks per instruciton single issue pipeline processor.

My modern installation of windows with my modern software has trouble fitting on my modern hard drive, nor does it run quickly on my modern processor.

Originally Posted by MK27

You get to learn how old-timers used to do it!

exactly one reason why i want to do this.
i have already taken the most simple programs and rewritten them and some of them a couple of times. wc as example i written more than one version of.

i have also off course still good up to date books like advanced porgraming in the unix environment and online books and forums to use.

i think it is great to have small programs to look at but i do however miss the option of using code from today standards but it seems to be bigger sources than i as beginner would start changing to much in.

but this way i learn to update others work and as far as this is only hobby of mine it is not a great skill to have but still a good learning process.

the thing is that i study unix systems as a complete personal interest and i will of course like to learn the tradition of programming for unix. in that context i do find these sources i linked to worth gold but i still would like to also get other input as well.

thanks for reading the post.

Originally Posted by cmay

i been under the impressing that only when writing on a kernel would ever need a goto in code and i also think from having read some other sources for the same programs from solaris and from gnu core-utils that these old codes seems to be not so well written and somehow very simple and i wonder if it is actual old production code.

Use of "goto" can sometimes be cleaner than introducing flag variables... but let's not start a huge anti-goto war here! There is nothing inherent in kernel code that would necessitate (or excuse) use of "goto" more than in any other context.

And you're right. Core-utils and such are not necessarily well written. I'm gonna get in trouble for saying this but in my opinion, early internals code was not written by seasoned mainframe computer experts but rather by learn-as-you-go hackers who started with hobby computers and had hardly a clue about advanced data structures and multi-threaded code.

I see crap all the time - for example, when examining some DNS domain resolving code which was supposed to be the type: "don't touch this: it's written by gurus" type code which was absolutely hideous. The code is in a production environment and expected to parse and search for 1000s of domain names a minute (second?). I had to get multi-colored markers out just to try and follow the umpteen-nested loops that spanned many pages. At least 10 colors.

Gurus. Yeah right.

I'm not sure of your exact question. I learned C from K&R. These days I wouldn't learn it from a super old non-ansi K&R (my first C book) but if you like K&R I wouldn't hesitate to use the later editions. In fact the only book I own specifically on C is K&R and the only book I own specifically on C++ is Stroustrup.

As far as gotos go, I have never used them much outside of very old BASIC and Fortran, but once in a while they may save you a couple of ifs. If you have a good reason, then use one. If you find some old code that helps you out, use that too.

Originally Posted by SyntaxError

I'm not sure of your exact question. I learned C from K&R. These days I wouldn't learn it from a super old non-ansi K&R (my first C book) but if you like K&R I wouldn't hesitate to use the later editions. In fact the only book I own specifically on C is K&R and the only book I own specifically on C++ is Stroustrup.

As far as gotos go, I have never used them much outside of very old BASIC and Fortran, but once in a while they may save you a couple of ifs. If you have a good reason, then use one. If you find some old code that helps you out, use that too.

i think nonoob answered the question.

i am learning c as a hobby and i study the old unix systems as i find them very interesting so what i do is i get some source code from books or online and i study those and then i try rewrite the code into my own version which is the todays standard c i use. i use the gcc std=99 flag when compiling on linux . and i use cc on open solaris for these old codes.

i think its a great way to learn about things that has a interest and i think to have a whole unix version 7 in source to study and modify and rewrite is very rewarding to me.

but i was sort of wondering if it really was old production code since some of it seems not very well written. i am very happy with the answers i got
thanks.