www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - GDC2, LDC2 Status Updates?

reply dsimcha <dsimcha yahoo.com> writes:
I've been curious for a while how close GDC2 and LDC2 are to being ready 
for production use.  Are the test suite results posted publicly for 
either of these?  Other than that, is there anything else other than 
building and testing for myself that would give me a good idea of where 
these libraries stand?
Apr 23 2011
next sibling parent Peter Alexander <peter.alexander.au gmail.com> writes:
On 24/04/11 2:14 AM, dsimcha wrote:

 I've been curious for a while how close GDC2 and LDC2 are to being ready
 for production use. Are the test suite results posted publicly for
 either of these? Other than that, is there anything else other than
 building and testing for myself that would give me a good idea of where
 these libraries stand?


I would also like to know the answer to these questions.
Apr 24 2011
prev sibling next sibling parent lurker <lurk lurking.org> writes:
dsimcha Wrote:


 I've been curious for a while how close GDC2 and LDC2 are to being ready 
 for production use.  Are the test suite results posted publicly for 
 either of these?  Other than that, is there anything else other than 
 building and testing for myself that would give me a good idea of where 
 these libraries stand?


The typical reply is "they're already fully production ready, it just depends
on what you need". Seriously though, people have different kind of requirements
for production ready products. These kind of questions just start yet another
public debate of the basic definitions of terms.
Apr 24 2011
prev sibling next sibling parent Andrew Wiley <wiley.andrew.j gmail.com> writes:
On Sat, Apr 23, 2011 at 8:14 PM, dsimcha <dsimcha yahoo.com> wrote:


 I've been curious for a while how close GDC2 and LDC2 are to being ready
 for production use.  Are the test suite results posted publicly for either
 of these?  Other than that, is there anything else other than building and
 testing for myself that would give me a good idea of where these libraries
 stand?


I know that Iain has been working through Brad's testsuite for DMD on 64 bit
GDC, but I don't think he has the results being generated in any automated
fashion. I believe 32 bit is pretty much production ready but 64 bit has
some issues, but he'd know that better than, well, anyone.
Apr 24 2011
prev sibling parent reply Iain Buclaw <ibuclaw ubuntu.com> writes:
== Quote from dsimcha (dsimcha yahoo.com)'s article

 I've been curious for a while how close GDC2 and LDC2 are to being ready
 for production use.  Are the test suite results posted publicly for
 either of these?  Other than that, is there anything else other than
 building and testing for myself that would give me a good idea of where
 these libraries stand?


I don't think it is "dsmicha's production" ready. Though if you write D2 code
conservatively there should be little reason why things should go wrong. :~)

Unittests that fail in the D2 testsuite can be found here:
https://bitbucket.org/goshawk/gdc/src/6e40c9c42f6e/testsuite/Makefile#cl-103

With 64bit, all the same tests pass with the exception to some vararg related
tests passing structs/unions.

However, the testsuite can really only been taken with a pinch of salt, and
certainly doesn't cover everything the language allows you to do.
Apr 24 2011
parent reply dsimcha <dsimcha yahoo.com> writes:
On 4/24/2011 4:04 PM, Iain Buclaw wrote:

 == Quote from dsimcha (dsimcha yahoo.com)'s article

 I've been curious for a while how close GDC2 and LDC2 are to being ready
 for production use.  Are the test suite results posted publicly for
 either of these?  Other than that, is there anything else other than
 building and testing for myself that would give me a good idea of where
 these libraries stand?


 I don't think it is "dsmicha's production" ready. Though if you write D2 code
 conservatively there should be little reason why things should go wrong. :~)


LOL.  I mean, this really made me laugh out loud.  I see I've gotten a 
(probably well-deserved) reputation around here for pushing compilers to 
the limit and finding every bug imaginable.  Yeah, std.parallelism is 
horribly broken with GDC on Windows.  I haven't tested on Linux yet.  As 
far as I can tell the brokenness is deterministic, so it's very unlikely 
to be latent concurrency bugs.  The funny thing is that most of the bugs 
I've filed against GDC have been culled from (of all things) a 
statistics library I wrote.  Same with most of the bugs I found in the 
DMD 64-bit port.  How a statistics library could possibly be so good for 
finding compiler bugs that the test suites miss, I'll never know.


 Unittests that fail in the D2 testsuite can be found here:
 https://bitbucket.org/goshawk/gdc/src/6e40c9c42f6e/testsuite/Makefile#cl-103

 With 64bit, all the same tests pass with the exception to some vararg related
 tests passing structs/unions.


This is the kind of thing I'm looking for.  The biggest reason I wanted 
this info is so I know what the known defects are and don't have to 
waste time reducing and filing bug reports for bugs where there are tons 
of probably related unit test failures already.

BTW, my statistics library now passes all its unit tests on 64-bit Linux 
as of last time I checked.  There's still one bug that only occurs on 
32, and it looks like a rather nightmarish cross-module bug in a 15,000 
line codebase.  I haven't gotten around to reducing/filing it yet.


 However, the testsuite can really only been taken with a pinch of salt, and
 certainly doesn't cover everything the language allows you to do.


Yeah, I believe it.
Apr 24 2011
next sibling parent reply bearophile <bearophileHUGS lycos.com> writes:
dsimcha:


 How a statistics library could possibly be so good for 
 finding compiler bugs that the test suites miss, I'll never know.


I have found several bugs even writing 20-lines long programs.

Bye,
bearophile
Apr 24 2011
parent reply Don <nospam nospam.com> writes:
bearophile wrote:

 dsimcha:
 

 How a statistics library could possibly be so good for 
 finding compiler bugs that the test suites miss, I'll never know.


 
 I have found several bugs even writing 20-lines long programs.
 
 Bye,
 bearophile


"Several", huh?

Top bug reporters:
bearophile: 482
David Simcha: 222.
Don: 216. (many of these are on behalf of other people).
Andrei: 173.
Thomas Kuehne: 155 (mostly on behalf of other people).
Max Samukha: 127
Matti Niemenmaa: 126
davidl: 115
Apr 25 2011
parent bearophile <bearophileHUGS lycos.com> writes:
Don:


 Top bug reporters:
 bearophile: 482


Some of them are enhancement requests and not true bugs... and recently some of
them are just Phobos bugs.

Bye,
bearophile
Apr 25 2011
prev sibling parent reply Iain Buclaw <ibuclaw ubuntu.com> writes:
== Quote from dsimcha (dsimcha yahoo.com)'s article

 On 4/24/2011 4:04 PM, Iain Buclaw wrote:

 == Quote from dsimcha (dsimcha yahoo.com)'s article

 I've been curious for a while how close GDC2 and LDC2 are to being ready
 for production use.  Are the test suite results posted publicly for
 either of these?  Other than that, is there anything else other than
 building and testing for myself that would give me a good idea of where
 these libraries stand?


 I don't think it is "dsmicha's production" ready. Though if you write D2 code
 conservatively there should be little reason why things should go wrong. :~)


 LOL.  I mean, this really made me laugh out loud.  I see I've gotten a
 (probably well-deserved) reputation around here for pushing compilers to
 the limit and finding every bug imaginable.  Yeah, std.parallelism is
 horribly broken with GDC on Windows.  I haven't tested on Linux yet.  As
 far as I can tell the brokenness is deterministic, so it's very unlikely
 to be latent concurrency bugs.  The funny thing is that most of the bugs
 I've filed against GDC have been culled from (of all things) a
 statistics library I wrote.  Same with most of the bugs I found in the
 DMD 64-bit port.  How a statistics library could possibly be so good for
 finding compiler bugs that the test suites miss, I'll never know.


As some anecdote goes, bugs will be found once you stop looking.
Apr 24 2011
parent Sean Cavanaugh <WorksOnMyMachine gmail.com> writes:
On 4/24/2011 3:51 PM, Iain Buclaw wrote:

 As some anecdote goes, bugs will be found once you stop looking.


Or when you want to show your app to someone else :)  I suspect this 
increases geometrically with the number of people watching and how many 
times you tell other people how cool it will be.
Apr 24 2011