The Big Programming Thread - Page 73

On August 03 2011 06:01 disco wrote:
This should work:
+ Show Spoiler +

On August 03 2011 08:56 Kahzaar wrote:

thanks, had to change some other smalls things too, but now it's working

On August 01 2011 11:12 EvanED wrote:

I vehemently disagree with this, for reasons I've stated above. (Just the negative part -- that you won't learn anything useful. Depending on your interest, Java or C++ might be a reasonable starting language.)

I've got a substantial portion of the "best" CS schools on my side too. US News gives the top 10 as MIT (#1), CMU (#2), Stanford and UIUC (#3-tie), Cal Berkeley (#5), Georgia Tech (#6), UT Austin (#7), Michigan (#8), Cal Tech (#9) and Cornell (#10).

Of those: MIT used to use Scheme and now uses Python in their intro class, CMU uses SML, Stanford uses Java, UIUC uses Java for what seems to be the primary first in-major class (there's another intro class that uses Python, but I'm not sure how it fits in), Cal Berkeley seems to emphasize Scheme though it's not at the complete exclusion of everything else, Georgia Tech uses Python, UT Austin uses Java, Michigan uses ??? (I actually can't find a course webpage, and the description doesn't say; there's a chance it uses C++. If you have better luck than me, the course is EECS 280), Caltech uses Python, Cornell uses Java.

Discounting Michigan, counting UIUC as Java, and counting Berkeley as Scheme, the counts are 4 schools for Java, 3 schools for Python, 1 school for Scheme, and one school for SML. Even if Michigan came into the picture with Java, that's still half of the US News top 10 schools which teach a language you accuse of teaching "nothing useful."

Now, the pressures are very different for a university deciding how to set its curriculum and an individual deciding what to learn, but I think that "you won't learn anything from a high-level language" is complete bogus.



Well, that's sort of true, but at the same time IMO a good C programmer does not necessarily make a good C++ programmer (though it is likely). You'll have to unlearn a fair bit of stuff for that.

On August 09 2011 20:58 Soier wrote:
For Android:
Since you need basic java to program for Android I would recomend getting the java skills before you start with your app. If you know basic java and use Eclipse with the Android dev kit you are well equiped på produce and test android apps.

Dont know much about iPhnoe programming, as far as I know you need to know about objective-C++ witch is a harder language to learn than java. Furthermore if you want you iPhone app on the market you have to jump allot of hoops and even pay for a lisence, so if you are just starting out I would recoment programming for Android.

Can PM me if you have further questions

On August 02 2011 20:55 Morfildur wrote:


Well, in my experience, 99% of the programmers shouldn't ever touch any code again because they just don't have any clue, so it's not suprising for me to see horrible code, even in government code from full time programmers :p

On August 14 2011 12:04 fabiano wrote:


What do you mean? o.O

On August 14 2011 13:23 Frigo wrote:
Sorry man, gets is inherently unsafe and vulnerable to buffer overflow attacks. Use something else, especially if it is C++.

On August 14 2011 14:06 Abductedonut wrote:If anyone can think of another solution using actual gets(), please let me know! I'm actually curious.

// global or file static:
jmp_buf from_fgets;

void return_to_past_gets(int ign) {
    (void) ign;
    longjmp(from_fgets, 1);
}

// Where you want to call fgets
struct sigaction old_handler;
char * buf;
int ret;
if (setjmp(&from_fgets) == 0) {
    // Set up memory. 'buf' will point to 4096 bytes, then a guard page
    buf = mmap(NULL, 4096*2, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
    if (buf == NULL) goto err_mmap;
    ret = mprotect(buf + 4096, 1, PROT_NONE);
    if (ret != 0) goto err_mprotect;

    // set up a handler for SIGSEGV, which will be 
    struct sigaction new_handler;
    new_handler.sa_handler = return_to_past_gets;
    ret = sigemptyset (&new_action.sa_mask);
    if (ret != 0) goto err_sigemptyset;
    new_action.sa_flags = 0;
    ret = sigaction(SIGSEGV, &new_action, &old_action);
    if (ret != 0) goto err_sigaction_new

    // call fgets
    fgets(buf);
}
else {
    // fgets overflowed.
    ret = ungetc(buf[4095], stdin);
    if (ret != 0) goto err_ungetc
    buf[4095] = 0;
}
ret = sigaction(SIGSEGV, &old_action, NULL);
if (ret != 0) goto err_sigaction_restore

// YOU implement err_mmap, err_mprotect, err_ungetc, err_sigemptyset,
// err_sigaction_new, and err_sigaction_restore

// Oh, and don't forget to
unmmap(buf, 4096*2);
// instead of free(buf)

fgets(buf, size, stdin);

On August 14 2011 14:49 Frigo wrote:
No sorry, the compiler is quite right about warning you, it simply can't be made safe. You need to know in advance whether the line fits in your buffer, and that is simply not possible with the standard input, only with files.

On August 18 2011 08:14 Craton wrote:
I have an interview coming up. What would be some good areas to brush up on beforehand? The job is primarily C#.NET and VB.NET in both client-side applications and ASP.NET.

Finished B.S. in May.