Fixes bug http://d.puremagic.com/issues/show_bug.cgi?id=5206 and introduces statvfs #257
Conversation
|
Thanks for contributing to druntime.
|
|
Hi! Tanks! Best regards, Robert |
ghost
assigned
|
@dawgfoto I'll assign this to you - please let @eskimor know how to proceed. Thanks! |
|
ping on this |
|
We try to keep this headers standard compliant. I think what we're currently porting is roughly SUSv3 (POSIX.1-2001 + XSI) which you'll get if @complexmath do you have any further information? http://pubs.opengroup.org/onlinepubs/009604499/nfindex.html This means __USE_XOPEN2K8, _BSD_SOURCE, _SVID_SOURCE, __USE_GNU, http://www.gnu.org/software/libc/manual/html_node/Feature-Test-Macros.html If __WORDSIZE is widely used in linux headers then it makes sense to declare it here. We should probably comment that these are global glibc/linux defines. I intend to start a test suite for druntime and would like to see your unit test in there, but I'm not so positive about ad-hoc unittest additions to the makefile. Please watch out to avoid tabs and trailing whitespace and use K&R style braces if it's the predominant style. There is also a small D style guide. |
|
|
||
| version(Alpha) { | ||
| enum __WORDSIZE=64; | ||
| } |
There was a problem hiding this comment.
Would it be possible to use D_LP64 instead?
There was a problem hiding this comment.
Well at least on x86 it seems to be the same. Thanks! I'll use it instead.
There was a problem hiding this comment.
__WORDSIZE should be determinable from D_LP64 on all relevant platforms these days.
|
Thanks very lot for your comments, I will examine your comments and proposals thoroughly. I am sorry for the delay as I was on holiday the last two weeks, but I just came back and I am motivated :-) unittests: absolutely. It is also pretty redundant work to add struct unit tests for every struct, unfortunately C has no mixins. But a code generator would do. If desired, I could write a pair of D mixin and C code generator for this purpose. Maybe having a directory for auto generated C unit test helper functions, which are automatically included in the unittest build? Should I remove the unit tests for now from the pull request? Or leave them until we have a better solution? Thanks for the style guidelines, I will adhere to them in the next update. |
No need to hurry.
I'd opt for leaving them out for now. I want to start a test suite like dmd and there would be plenty of space for such tests, so it's probably even OK to write and extend them manually. |
| enum __WORDSIZE=32; | ||
| } | ||
|
|
||
| version(MIPS6) { |
|
Alrighty, waiting for one week on this. |
Avoided unrelated whitspace changes. Removed unit tests, as they are being replaced by a more general suite, according to Martin Nowak. Work in progress. Not yet tested.
Added version(posix) section in config, currently not used. Fixed whitespace in config.
As suggested by Martin Nowak.
|
Just updated request. It now only contains the needed changes to make struct stat work. Although the unit tests are now no longer included in the request, I tested the commit with them and they worked with both MODEL=32 and MODEL=64. In addition I tested my deimos-fuse wrapper with it and it seems to work too. |
|
I merged this with some smaller adjustments ccaa0ef, thanks. |
Hi!
This little patch fixes afore mentioned bug, that the stat struct is not correct on Linux systems.
It also introduces a unit test for testing the correctness. In addition I added the statvfs struct for Linux systems, also with unit test.