Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Moghjubar]

Speaking of !fun! dangerous things, heres a nice talk on compiler tuning and benchmarking: https://www.youtube.com/watch?v=nXaxk27zwlk 

Will make you appreciate compilers and the lower end of programming.

[Reelya]

@Telgin: Actually, you have some details backwards about dynamic memory and the stack vs heap.

Heap = dynamic, i.e. things created with new and delete. Dynamic things don't call their destructor, unless you tell them.

Stack = non-dynamic. i.e. things created as normal variables in functions or scope blocks. They automatically call their destructores when the scope ends.

~~~

So, if you made a regular string or had a string as a parameter for that function, it's likely that the string was packed in memory next to the array, and you over-wrote part of the class. Then, when the function ended, all local variables have their destructors called when the stack frame is cleared, and this would have caused the error. Actually, if you had made it a dynamic string that error would never have occured (though you'd have a memory leak instead).

[Telgin]

Yeah, that's exactly what happened I guess.

I guess I shouldn't say I was surprised that it was crashing in the string's destructor despite the string being stack allocated.  I was more surprised that it was crashing at the start of a function invocation that took a string as a parameter, which I was constructing at that moment.  So, the string didn't even exist at the point that I was trashing the array.  I guess the compiler was reserving stack space for the string before that point or something.

Anyway, the behavior is most surprising since internally strings use heap memory anyway as far as I know.  I suspect that I was overwriting an internal pointer to the memory block, so when it tried to free the memory in its destructor it freed the wrong pointer.

Regardless, it sucks that Valgrind can't detect stack memory errors.  Well, its default Memcheck tool can't anyway, but I think one of the others can.  I should probably read up on that at some point.

[3man75]

Not so much a programming question but where would be a good place to keep all of my programming projects for possible employers/curious individuals to see?

[Reelya]

Make a blog, and link screenshots + downloads from the blog. Then on the front page of the blog you can link your best projects as stickies, with a one/two-sentence description. Then you only need to provide the blog link, and interested parties have the option of browsing more content if they want to.

[Antsan]

Not so much a programming question but where would be a good place to keep all of my programming projects for possible employers/curious individuals to see?

I use bitbucket for my repositories, which can be linked to, and I have a StackOverflowCareers profile, where I can link these projects at a central point.

[RoguelikeRazuka]

What are friend classes (C++) useful for? For example, there are two classes A and B, and the A class is a friend of B. How can one apply this?

[Reelya]

Well, if you want access to things in a class normally you have to make them public, or make public getters and setters. So they become part of the global interface for that class. But say you want to project a more limited public interface, because it's part of a library or something, yet there are some support objects that need to have direct acess to stuff inside the class. Using the friend keyword means that the support classes can do their jobs, but without exposing more details in the public interface of the class in question.

Also, since the "friend" system is a whiltelist, you can narrow down which other classes might have made changes to the object in question, so it might reduce overall program complexity and make it easier to track down bugs.

[jaked122]

Not so much a programming question but where would be a good place to keep all of my programming projects for possible employers/curious individuals to see?

I've heard that it is pretty much industry standard to examine github pages. Those are generally better looking and have the upside of not requiring a curious programmer to deal with archives if they have git installed. Which they should.

Git is the best.

I should make a tutorial on git actually. That could be useful here.

[miauw62]

Yeah, git is great once you get used to it and I recommend anybody who can't use it to learn to use it.

[jaked122]

Yeah, git is great once you get used to it and I recommend anybody who can't use it to learn to use it.

It's been so long since I started using it that I just know how to look everything up. I can't remember having issues. It's a lot better than svn. That's not hard. Fossil is better than SVN, as is DARCS, Mercurial(hg) or any other distributed scm.

Also github just looks good. That's why I prefer it to the alternatives like bitbucket or sourceforge(do they still host code?).

[Orange Wizard]

To quote the CEO of Microsoft, everyone uses git and it's cool.

[3man75]

To quote the CEO of Microsoft, everyone uses git and it's cool.

Microsoft. You may not like our products but they work.

EDIT: IMO on how I feel about their stuff.

[Antsan]

To quote the CEO of Microsoft, everyone uses git and it's cool.

Microsoft. You may not like our products but they work.

EDIT: IMO on how I feel about their stuff.

What has that to do with git? Wikipedia says

Git was initially designed and developed by Linus Torvalds for Linux kernel development in 2005.

I prefer Mercurial. I don't really get what git's advantage is supposed to be. It just seems harder to use without offering anything useful in return.

[jaked122]

I just don't see the difference after this long.

Spoiler: wow minitutorial (click to show/hide)

Let me try to explain git. I don't know what is difficult for you, but it works fairly simply. Especially after knowing a bit of the internals.

So let me  start that git is a sqlite database with a host of applications that attach and manipulate it. This database lives in the .git directory.

Don't go there. It is not for us mortals.

This database is clever and fast and very adept at managing changes between versions of files, but it doesn't monitor what you've done in the repo other than when you invoke the various git utilities.

The base git command is a bash script essentially that delegates to other tools in its directory or something. It doesn't matter.

Code: [Select]

git init #starts a repo
Creates the empty database.

Code: [Select]

git add <file>
adds the instance of a file to the git. This does not change the repo's state until you commit.

Code: [Select]

git commit
Creates a commit with additions and changes, git sees files in terms of modifications made to them. Lines added and removed. The file handle can be changed with another git commit.

Code: [Select]

git mv <file-src> <file-dst>
This changes the file name associated with the changeset.

Code: [Select]

git rm
This removes a file. Have you noticed a pattern? These are unix shell commands.

Code: [Select]

git branch
Changes the active branch of the git repo. The active branch is the active changeset being worked upon. These start from a single commit, and then move onwards independent of each other.
Once you have done this, then you must checkout the new branch.

Code: [Select]

git checkout <branch-name>
This changes branches.  You can also switch between remote branches this way if they exist on the local repo as well.

Code: [Select]

git merge <branch>
This merges the chosen branch into the current one, this closes the other branch.

Code: [Select]

git pull <remote>
This pulls changes from the remote.

Code: [Select]

git push <remote> <branch>
Push the state of the repo to a remote.

Code: [Select]

git gui
This is in most distributions of git. If you can't use this, you don't actually understand how scms work, or at least distributed scms. In this case, one might term you one of the following:

• A scrub
• A noob
• I missed your problem with it.

There are many more. But fortunately, there are man pages for these, and synopsis online, and tutorial videos, and textual tutorial. All that matters is that you understand that if you want your repo to be publicized, at the moment it is simpler to do it with git.