string
i dunno
i guess i should use sprintf huh/

http://www.pennergame.de/ref.php?uid=348 klick plx

hmm
i certainly would.

ok

what is mode +d?

realname ban

oh

buondi'

bah that didnt fix anything

sawatdi khrap

Morning dudes.

hi, i have a question
how can an array be initialized in constant time?

int a[10] = { 0 }; will initialize a in constant time.
it's very hard not to do so in constant time, since the dimension of an array is constant in C.

not that there are other forms for initialization.

but, it might be that you have not asked the question that you think that you are asking.

i'm looking at the output of cpp [the c pre preprocessor for gcc]
and i have lines like
# 33 "neo_my.c"
in it
is that valid ?
valid c i mean

compile it with -pedantic -Wall and you'll see

the C compiler will understand whatever its companion CPP produced

seems to be a GCC extension
so i guess it's alright

gcc uses it to resolve error messages to included headers.

ok, good to know

Hello.

Hey

One simple question about syntax

yo
shoot

Just two if/else, one works, the other one not
http://rafb.net/p/ScgXJb87.html

Archville pasted "Why ?" at http://rafb.net/p/ScgXJb87.html (11 lines)

heh

the second statement
wait…
that's a homework question

hello; is there a channel for questions related to autoconf and automake ?

##workingset

thanks

no homework, i'm just bad at syntax sometimes

you're lacking closing braces

consider which if statement the else is responding to

even in the first one

Chris yes, that was my fault at pasting on the pastebin
(on the first one)

then fix the pasting

Ok, sorry, let me fix it

use a code editors automatic indentation
that'll help

i don't like automatic identation

it'll help in this case
help you to understand why the first one works and the second one doesn't

http://rafb.net/p/HFaKBR14.html (only added braces at the end of the first one)

Archville pasted "Why ?" at http://rafb.net/p/HFaKBR14.html (11 lines)

I mean… both "if" and "else if" only receive one parameter

nothing to do with that

why do i need then to use curly braces ?

which if() does the else relate to in the first? then which one does it relate to in the second?

Uhmm
Oh
I got it

the syntax is fine

it's "legal"

yes
hence "fine"

it's only "fine" if it does what you want it to do

Thanks for the tip

yes, but he said "doesn't work" without providing behaviour example, so I can only say the syntax is legal.

p = malloc(4) is "legal" but not "fine"

I would provide an example

sometimes you have to read between the lines

why is that not fine?

but it's kinda complicated
i mean, it's a part of a 7.000 lines program

then simplify it

no, Archville needs to learn to be more specific than "doesn't work"

he can learn in time

hopefully

Yes, "doesn't work" is a very bad description

for windows, it's adequate.

simple fact is, i could see just by reading his code what his problem was, so "doesn't work" was adequate for me

but explaining what it should do… it's kinda hard here without the rest of the program

what is not fine about p = malloc(4)?

why 4
use char p[4] in that case
(for great simplification)

what are you talking about?

p=malloc(4)

it was an isolated example scott_w gave of something that wasn't "fine".

i never finished
it wouldn't be "fine" if you wanted to malloc the size of an integer for example

oh, okay.

as it wouldn't be portable

right

are there any standard libraries that define a byte type?

What do you mean?

e.g. typedef unsigned char Byte

Ugh

a char IS a byte.

i know

the word byte appears in the standard many times but not as a type(def). there being no other standard libraries, no.

if a byte is 8bit, char is at least a byte, but may be bigger…

but it doesn't look nice when you write it

sizeof(char) is always 1.

char is defined to be a byte

char IS a byte, and a byte may be more than 8 bits.
if you want 8 bits, see if the implementation has uint8_t

an unsigned char must encompase all of the bits of a byte.

just thought that it would save me 7 seconds of defining it myself
not a big deal really

whether char or signed char do too depends on the implementation.

you just wasted more than the 20seconds it would take asking if it was already done…

7 seconds per program as i'd have to #include the header that defines it

why bother? unsigned char isn't so much to type. and if you mean an octet then it suffices, even if it is wider.

that's why typedefs are bad (for structs) - you have to #include lots of files

just for readability

and that explodes compile time on some projects

makes the variable's use explicit
Byte a; is clearer than char a; for example

no it isn't

it is not

well, ok, so do that.
whatever you do, you'd have to include a header for it, so stop whining.

:P
true

personally I prefer byte_t

as a programmer who does not know your program, I'd have to lookup Byte first
so unsigned char a; will do best.

char is a language defined type, you don't get clearer than that in your code.

but only if it really needs to be unsigned

sByte and uByte if that makes you happy

just use char…

and less camel-case.

i tend to avoid _t's, posix having bitten me in the past. even c came close with c99's fexcept_t was close, and i shan't want to pray about futures.

what was fexcept_t typedefed to?

char is problematic

damn, i need foor
food*

int a[10] = { 0 }; will initialize a in constant time.

i did not mean it this way, what if it's not 0 and it's different values, i read there's some complex data structure that can be used to initiliaze it such way and i was wondering if you know about it

int a[10] = { [0 … 4] = 1, [5 … 9] = 2 }; ?

is that standard c or some funky gcc extension?

some funky extension, i think
but not gcc afaik

it's certainly not standard C

ranges are a gcc extension
works also for case 1 … 100: e.g.

they are also recognized in ICC

oh, hi

hm?

I did not know you were here

I don't even know you

I know you from the LKML

well if that's all - there were not any discussions

I can't follow you

there has been no mail thread in which both you and me were involved

well, i meant that i had an except_t and fexcept_t is fairly close. as for what it is, that's up to the implementation — probably an unsigned something — consider it a magic cookie.

-that I can find.
you could have written struct except instead

indeed, except the project manglement lubbed typedefs.

well even if they love typedefs, you don't need to add any

sigh.

http://lkml.org/lkml/2006/11/21/34
fwiw

nothing, most likely.

there was one some months ago.. was about coding style, especially indentation & alignment

whatever

is there an obvious reason why a program could get SIGQUIT when calling tcsetattr?

you received a quit signal due to the changed VQUIT?

Twi, what is VQUIT?

you managed to call tcsetattr but you didn't find the documentation for VQUIT?

key sequence to trigger quit

Twi, I'm debugging diald

a) i'm not Twi; b) do you have any documentation for tscetattr? and c) are you familiar with c?

int a[10] = { [0 … 4] = 1, [5 … 9] = 2 }; that's still not o(c), because the real implementation of this array is done with o(n), anyway, the answer isnt as simple as that

twkm, I have the manpage, I am familiar with c

then you should look on the man page or one in the see also section for a description of VQUIT.

most implementations have it O(n)

my guess is that you called tcsetattr without having first used tcgetattr.
so now some random (perhaps nul) byte is being interpreted as the quit character.

twkm, it calls tcgetattr, sets the whole c_cc array to 0, then sets VMIN to 1
the sigquit occurs only sometimes

err, i get a problem with ((Tcl_Obj **)objv) += 3;
Tcl_Obj *CONST objv[]
i don't know what ((Tcl_Obj **)objv) += 3; actually does

good.
it is an "lvalue cast", which is not allowed.

yah

so don't do that.

so i want to know what's the old behavior

old?

it should be compilable in older GCC

the *valid* behavior is that casts don't result in lvalues.

it's current insight GDB code
tcl code actually
i fail to compile it

and you are writing this code?

i don't know what magic makes me compiled it yesterday, possibly because of my old GCC
no

if you have problems with some package you should seek the help of the maintainer.

AFAIK GCC 3.4.5 compiles it well X_X

gcc sucks.

heh

uninstall gcc 4, reinstall gcc 3.

you recommends which gcc?
*compiler
which compile hosting is better ?

i don't recommend any compiler over another. (i curse each of them equally, usually)

lol @ twkm
heh, it's funny compiler couldn't implement simple C
well

well, the previous statement was not valid c, so …

so compiler should have already caught that bug
now the compiler just leave the undefined behavior there

recent versions of gcc removed their idiot lvalue casts, so now it is a normal error and prevents compilation. have the maintainer remove the horribly unportable construct and the problem goes away.

you see, it's trouble-some to get maintainers of insight informed

sorry to hear it.

or maybe tcl's
i haven't tried
just think it would be troublesome
tclWinDde.c:1178: warning: use of cast expressions as lvalues is deprecated
so what doesn't that warning mean?
it won't generate code for it?

it looks like code will be generated.

pff
gcc 3.4.5 compiles it

what it means that soon your compile will consider it an error, and instead of going forward it will croak there instead.

hey, my code is here, http://rafb.net/p/RfvIcF89.html in function printText, it prints to the console correctly the string, but in the glWindow, it only prints out a single character. however if i comment out calls to printText and textParser, and uncomment the block comment, everything else prints correctly, can anone help?

not really, but please do tell us when you find it!

wtf
i stripped it into a small app
gcc 3.4.5 doesn't compile it any more
it emits me test.c:42: error: invalid lvalue in assignment

DavidLeon, you have an error in your code then; if you paste it I can have a look.

it's the problem i discussed with twkm a while ago, it's invalid code, though i need to know how it was compiled by gcc

I'm confused
if your source code is invalid C then GCC has nothing to do with it
any compiler would give you the same error, no?

it gives me warning in windde.c
now i stripped the code to the minimum case, it gives me an error

can I see that minimum case?

wait
this time gcc gives me even more funnier error messages

how about you paste them?

*even funnier

paste

http://rafb.net/paste

of course i would paste

well then quit stalling and paste it already

http://rafb.net/p/Da8UNK67.html

there's a good chance it's just a silly syntax or simple semantic error, and we'll spot it right away
line 42 you say?

it's undefined?
huh

you said before that you were getting an error with test.c:42 as the line spec
is this that file?
oh, wait
nm
I see, you put the errors at the bottom

it's 47

yea

the error message is damned weird..

I think the error is on line 40
DavidLeon, the type "ClientData" isn't defined anywhere

aye

GCC doesn't know what to do with it, so it just sees it as a syntax error

why does everyone hate goto?

It can easily lead to unmanageble code.

syntax errors will often swap garbage errors further down the line, that's why the error on line 47 (which probably will go away if you fix the error on 40) looks so weird

how
?

http://xkcd.com/292/ == why gotos are bad

Because Djikstra thinks it's bad.

gotos are not always bad
stupid people bans smart people of using it

I agree with you.

*ban

http://c-faq.com/style/stylewars.html

thanks

actually i find goto statements are useful sometimes

To me they've only been useful in deeply nested loops

to me only for error recovery

arkx no
not really
gotos are not that restricted
consider you get a great bunches comparisons
and all comparisions goes to a cleanup after the conds hit
if not hit, then go on
the comparision
their, goto is better than any stupid continue or something

paste an example
'there'

yah *there

hiya there

if (strchr(foo,bar)){dosomething(); goto cleanup } // bunches of them
if the test not passes, then go on another test

where's the test?

test refers to strchr(foo,bar)

why not use else if

O_O
it's not only testing 1 thing

you need to paste a better example

it's going to test foo for few , and anotherfoo for another few

if (strchr(foo,bar)) {} else if (strchr(foo2,bar2)) {}

no
wait a little bit
yes
it can be rewritten as else if chain
though imo , goto is cleaner
and the aim of writing code is for cleaner looking
not just simply avoiding using goto

goto's are -fine-, much like guns.
(ie, don't start off learning to code with em. Unless you're writing assembler rather than C.)

gcc 3.4.5 doesn't give me error on this :http://rafb.net/p/yR6tpa24.html

I can see how they are like guns that respect

it only throws me warning
on invalid code

y.c:47: error: invalid lvalue in assignment

you're not using gcc 3.4.5

of course not

so that's the point

so rewrite it
looks like gcc 345 bug

i need to know the old behavior first

Tcl_Obj **foo = (void *)objv; foo += (async+3);

i would figure it out first
no

that is what the broken line reads to me
plus, objv=op

your code won't take effect over objv

that I added

objv = op?

objv = (void *)op
but yes.
I find it ambiguous
since (type)object is not an lvalue for me
at best - compile it with 3.4.5 and look at the assembler output

actually what i read the code should behave something like : Tcl_Obj *foo = *objv; foo+=3
yep, that's what i prepared to do

not to me, since there you += the (Tcl_Obj **)
and not the Tcl_Obj *
argh
it's just highly confusing

be careful with my code

if that's how it should behave, write it that way.

i just plain guess, i need to view the assembly

*shrug*

ppl keep showing their interest so i show my code and test results to them

either works, since sizeof(Tcl_Obj *) usually == sizeof(Tcl_Obj **)
which is why ++argv; does the right thing as much as (*argv + 1)
err
just fix it

i see lots of people put horrible casts into their qsort comaparison functions, just to avoid a temporary that the compiler will likely not even create if optimization is enabled. a false economy, usually.

yah

the same thing drove the lvalue cast. "hey, i have a not-right pointer but need right behavior, and wow it would be way too time consuming and ugly to open a block for a temporary", so bingo the arrival of lvalue casts.

yeah , as i expected

i see the same thing often in qsort comparison functions. dunno why but people seem to love to cast. sadly far too many do it poorly.

http://rafb.net/p/9FrWsc13.html
anyway i suggest ppl not code in C
C is error-prone

People is error prone, not C.

people are error-prone. the language cannot fix that. (what, you remove casts and that fixes it? hah!)

it's weird the code gives me magic
the pointer to an array is not get changed after the func returns
well, anyone want to write an equivalent to http://rafb.net/p/9FrWsc13.html

Tried a different compilator?
compiler*

i find i couldn't write something equivalent to http://rafb.net/p/9FrWsc13.html X_X
i mean line 47 equivalent

and what is that supposed to do?

it prints the result in my paste bottom
and it behaves as i expected

why do you even need to cast objv?
it is already of the proper type

it's not i want to cast
the coder write that piece of shit
did that cast

well in this recent paste, you can drop the lvalue cast

X_X
i didn't notice it

int *CONST objv[] — minus the const that is equal to int **objv

isn't it already an int ** ?

though what if it's not int* CONST objv[]

well then figure

i take int* CONST objv[] to simplify the test

change the type of objv

yes changed it into something else
it behaves almost the same as int* CONST objv[] case
i mean principlely
ok, it doesn't need a cast seems
O_O
thx u guys help
bye

const char *doc_root; const char *script_name;
const char *script_filename=strcat(doc_root, script_name);

doc_root doesn't point to any allocated space
neither does script_name
that code will kill your family

will post the full source to rafb.. mom

I don't have to fiddle with header files any more. I'm gettin spoiled, lol.

http://rafb.net/p/ivz6s796.html
thanks Cin for taking a look at this!

goodness, that's one brave programmer

i don't do a lot of things with c normaly but this time it has to be

tell quest about paste descriptions

use meaningful descriptions in your pastes

bah, no meaningful description so we get two lines anyway.

hehe in fact the purpose is that we have mod_fcgi working in a chroot environment and need to truncate the php filename so that it works in the chroot
uhehe in fact the purpose is that we have mod_fcgi working in a chroot environment and need to truncate the php filename so that it works in the chroot/u

how does Fortran compare to C in practice? I've been studying Fortran 90, and I'm curious.

heh, I just wrote my tell command so I thought I'd try it

so, what don't you understand about Cin's response?

I've been studying materials from gfortran (latest version of gcc 4.2.1)… and once you learn about the extensions to Fortran, both gnu and f2003, it's a capable enough language for many practical needs.

consider carefully the purpose of the qualifier.

There are some unigue issues hard to overcome using Fortran; but negligable for the time being.

i do understand the response but that does not bring me any further as in the full source both char* are correctly set

imaginator, have for f-code you'd like to share?
s/for/some

do you have a question regarding that paste?

what does the qualifier you use with stript_filename mean?

i thought that strcat (*a, *b) should return a string a+b

gulp!

not really. I've just been going through this tutorial in my spare time: http://www.pcc.qub.ac.uk/tec/courses/f90/ohp/header_ohMIF_1.html

would you like the official gfortran documentation?

you thought wrongly.

you mean the *script_filename?

no.

http://gcc.gnu.org/onlinedocs/, and http://gfortran.info/, http://gcc.gnu.org/wiki/GFortranStandards are the main ones

i'd imagine that a fortran channel might have further insights.

#fortran is available, yes.
and my channel #ModernCalcs is also available.

okay, but byside the qualifier there must be something else wrong, any hint?

but there is valuable GCC docs there as well in that /onlinedocs/ url

that seems sufficient for now.

one thing that bothers me about what I've read about Fortran90 is that it is possible in some cases to get undefined behavior when using allocate. So it's like the malloc and free issues of C

could you point me to a good help page for my problem?

i don't know of any good learning c "help" pages. i could point you at some books … http://www.iso-9899.info/wiki/Books

so while malloc and free work well generally when used properly. I still find I'm more productive in a language that has reference counting or GC, especially when dealing with lists.
Also, in callbacks for some code I've found problems with callbacks freeing the underlying data structures that called them, so when the functions unwind (an obvious design problem) faults can occur.

thx twkm!

My solution to that problem was to tag each struct when in a certain state, and then invoke a reaper to process the data structure and release any data at a safe time, after unwinding
Not ideal, but it works and it's fast and reliable. Before that the code was relying on a reference count in the struct, and that wasn't an ideal situation for quality reasons.

considering that arrays can be reshaped using the new Fortran extensions, malloc is rarely required

oh I wasn't aware of that. That sounds better than allocate

read through this when you get time. It's up to date: http://gcc.gnu.org/onlinedocs/gcc-4.2.1/gfortran.pdf
read through this when you get time. It's up to date: a href="http://gcc.gnu.org/onlinedocs/gcc-4.2.1/gfortran.pdf"http://gcc.gnu.org/onlinedocs/gcc-4.2.1/gfortran.pdf/a