Development mailing list

On Sun, Mar 14, 2010 at 3:25 PM, Ingo Weinhold ingo_weinhold@xxxxxx wrote: The boot loader is still compiled with -fno-rtti. Though I don't think there's a particularly good reason why that is. By not using RTTI, the boot loader is a little smaller than it otherwise would be. That could be important, because of all the weird space constraints that the boot loader has to deal with. Roughly the way RTTI works is that every class that has at least one virtual function has a static pointer (a class member variable) that ...

On 2010-03-14 at 20:31:10 [+0100], Axel Drfler axeld@xxxxxxxxxxxxxxxx wrote: Brecht Machiels brecht@xxxxxxxxxxx wrote: Maybe I incorrectly made this assumption seeing an -fno-rtti in some FS code. That's very well possible. The boot loader is still compiled with -fno-rtti. Though I don't think ...

Brecht Machiels brecht@xxxxxxxxxxx wrote: Is there any sort of documentation for these classes? No, but they are pretty easy to use, just look at an example usage in the kernel. It's not clear to me (but I haven't read the code thoroughly) why the stored element needs to be aware of the container, as the STL containers do not require this. I'm sure there's a good reason for it though :) ...

At Sunday, 14-03-2010 on 15:01 Ingo Weinhold wrote: BTW, using the STL (beyond stuff that is fully defined in the headers) for an FS in not such a good idea, anyway, since you won't be able to use it in the kernel. Are there any replacements for the STL list, vector and map classes? IIRC, other filesystems in the Haiku tree use custom container classes. ...

On 2010-03-14 at 14:30:35 [+0100], Brecht Machiels brecht@xxxxxxxxxxx wrote: At Saturday, 13-03-2010 on 21:21 Ingo Weinhold wrote: I believe there's a using namespace std; somewhere which causes this problem. I don't think there's anything one can do besides avoiding to include both headers in the same compilation unit. And this using namespace std; cannot be removed because of backward compatibility, I assume? ...

On Sun, Mar 14, 2010 at 8:30 AM, Brecht Machiels brecht@xxxxxxxxxxx wrote: Are there any replacements for the STL list, vector and map classes? IIRC, other filesystems in the Haiku tree use custom container classes. I suppose I can use BString instead of std::string. If you're intending to write this solely as a userland filesystem then using BString is fine. If you want it to be able to live in kernel though, then nothing in libbe is allowed as a kernel add-on is not allowed to link to libraries other than the kernel itself. This is also another reason why the STL is out (kernel-side), as that requires ...

At Saturday, 13-03-2010 on 21:21 Ingo Weinhold wrote: I believe there's a using namespace std; somewhere which causes this problem. I don't think there's anything one can do besides avoiding to include both headers in the same compilation unit. And this using namespace std; cannot be removed because of backward compatibility, I assume? Why is this only a problem with gcc2? BTW, using the STL (beyond stuff that is fully defined in the headers) for an ...

Great! Thanks Ingo On Mar 13, 2010, at 1:08 PM, Ingo Weinhold ingo_weinhold@xxxxxx wrote: On 2010-03-13 at 21:40:50 [+0100], Brian Landsberger ...

On 2010-03-13 at 21:40:50 [+0100], Brian Landsberger brian@xxxxxxxxxxxxxxx wrote: Franois Axel, This week I studied the existing NFS implementation as well as the bug (5496) that Michael Franz reported (re: Idiots Guide NFS on Haiku). I was also able to compile the userspace client to run it successfully on Haiku, although I must admit that the performance is a bit less than I had hoped for. As for next steps, I still need to ...

Franois Axel, This week I studied the existing NFS implementation as well as the bug (5496) that Michael Franz reported (re: Idiots Guide NFS on Haiku). I was also able to compile the userspace client to run it successfully on Haiku, although I must admit that the performance is a bit less than I had hoped for. As for next steps, I still need to study the custom network fs that Ingo wrote, and after looking into the current NFS code I am still left with some basic questions ...

On 2010-03-13 at 19:49:16 [+0100], Brecht Machiels brecht@xxxxxxxxxxx wrote: I'm getting stuck building assignfs with GCC 2.95. It builds fine with GCC 4. In file included from /Data/code/haiku/headers/cpp/list:34, from /Data/code/haiku/src/add-ons/kernel/file_systems/assignfs/A ssign.h:10, from ...

Hi, I'm getting stuck building assignfs with GCC 2.95. It builds fine with GCC 4. In file included from /Data/code/haiku/headers/cpp/list:34, from /Data/code/haiku/src/add-ons/kernel/file_systems/assignfs/Assign.h:10, from /Data/code/haiku/src/add-ons/kernel/file_systems/assignfs/Volume.h:24, from ...

Ingo Weinhold wrote: There are several casual contributors and developers who don't (yet) have commit access. I believe it would benefit the project, if we simplified things for them and also simplified the patch related workflow. E.g. it would be nice, if Trac could present an overview of pending patches, help with reviewing them, and maybe even provide a Commit (and Test build and commit :-)) button. ...

On Fri, Mar 12, 2010 at 11:13 AM, Ingo Weinhold ingo_weinhold@xxxxxx wrote: There are several casual contributors and developers who don't (yet) have commit access. I believe it would benefit the project, if we simplified things for them and also simplified the patch related workflow. E.g. it would be nice, if Trac could present an overview of pending patches, help with reviewing them, and maybe even provide a Commit (and Test build and commit :-)) button. WebKit pretty much has this exact process already set up. They use ...

On 2010-03-12 at 16:34:28 [+0100], John Scipione jscipione@xxxxxxxxx wrote: I would like to see the Haiku project move from subversion to a distributed version control system such as git or mercurial for purely selfish reasons. Distributed version control systems allow for a more laissez faire approach to source control which is better for me (if not better for the project). [...] ...

Hello John, On 12 March 2010 16:34, John Scipione jscipione@xxxxxxxxx wrote: I am unlikely to be granted svn commit access nor do I desire it as I am far from a mainline developer. Consequently, the code that I write gets added to my local copy only. When I perform an svn update my changes sometimes conflict with other people's changes creating a maintenance chore for me. Git and mercurial include features to make merging easier and avoid conflicts better than subversion. ...

On Mar 12, 2010, at 10:47 AM, PulkoMandy pulkomandy@xxxxxxxxx wrote: However, if there are others out there like me (perhaps with better changes) then subversion really is hurting the project by hiding good code in peoples local branches rather than allowing them to put those changes on the Internet. ...

However, if there are others out there like me (perhaps with better changes) then subversion really is hurting the project by hiding good code in peoples local branches rather than allowing them to put those changes on the Internet. I feel it's more desirable to have good, finished patches all available on trac for inclusion on the svn, than a lot of hackish, half-finished, outdated and incertain patches floating all around the internet. ...

I would like to see the Haiku project move from subversion to a distributed version control system such as git or mercurial for purely selfish reasons. Distributed version control systems allow for a more laissez faire approach to source control which is better for me (if not better for the project). I am unlikely to be granted svn commit access nor do I desire it as I am far from a mainline developer. Consequently, the code that I write gets added to my local copy only. When I perform an svn update my changes sometimes conflict with other people's changes creating a ...

2010/3/12 Varrun Ramani varrunr@xxxxxxxxx: Hi I have created a patch for Ticket #3618. I have uploaded the patch file also. What is the process for getting it integrated into repo? This is my first time contributing, so can i know the process for getting it integrated into the CVS? Hello, * The process is to attach the patch to the ticket and wait for the ...