OLPC - no problem, it's standard linux.
iOS - depends on the boost libraries used, but probably gonna make it a 
no go for a while.
Android - no chance. Not really good for C++ apps @ the mo anyway.

i hear what you & jon clark are saying about Boost. I have similar 
concerns about using up-to-date automake. However, don't like a 
situation where we have to write and test things twice - with and 
without boost versions, single & multi-threaded etc.

baz has a point about retaining single-thread in though. For 50 
sentences with hiero model (1000 pop-limit), time taken was 218sec 
(multi-threaded, 1 thread) v. 192sec (single-threaded), ie. 
multi-threading overhead makes decoding 10% slower

On 22/09/2011 20:44, Lane Schwartz wrote:
> I typically compile with Boost. That said, I do have a couple of
> (relatively minor) concerns.
>
> Would requiring Boost complicate/prevent the compiling and use of
> Moses on smaller mobile platforms, such as iOS, Android, and OLPC?
> When I was working with Hieu on the iPhone port, I don't remember
> having to compile Boost. It probably is possible, but I don't know how
> it would affect compiling on iOS, if at all.
>
> Compiling against Boost can sometimes be a pain, specifically in the
> case where you're using an older Linux distro that doesn't ship with a
> modern enough version of Boost. Unfortunately a bunch of the machines
> in my lab are still running CentOS 5.x, which has an ancient version
> of Boost. I have a newer version of Boost compiled on the machines,
> but I've definitely had problems convincing autotools to find and use
> that version when there's also a version installed in /usr. So to
> avoid that pain, when I'm on a machine with an older Linux distro, I
> typically just compile with Boost turned off.
>
> So as much as I hate to be the kill-joy, if Boost is made mandatory,
> please require the oldest version that provides the needed
> functionality.
>
> Cheers,
> Lane
>
>
> On Thu, Sep 22, 2011 at 8:24 AM, Christian Hardmeier<[email protected]>  wrote:
>> Hi,
>>
>> I very much agree with Barry when it comes to Boost. In my opinion it should 
>> be required, but there should be some control over the packages used. While 
>> I can't see a good reason not to use smart pointers, you probably don't want 
>> people to start using lambda expressions all over the place, especially when 
>> some of the developers aren't familiar with this kind of code.
>>
>> Cheers,
>> Christian
>>
>> On Sep 22, 2011, at 2:11 PM, Barry Haddow wrote:
>>
>>> Hi
>>>
>>> Here's my thoughts:
>>>
>>> - there should be single and multi-thread compile paths so single-thread 
>>> users don't pay the lock penalty. Maybe a -threads 0 works, but then you 
>>> have to check a config each time you want to lock
>>> - boost  should be required, but care about which packages we use
>>> - cruise control will help catch these compile path errors, once it's up 
>>> and running
>>>
>>> Cheers Barry
>>>
>>> Sent from my ZX81
>>>
>>> ----- Reply message -----
>>> From: "Miles Osborne"<[email protected]>
>>> Date: Thu, Sep 22, 2011 11:33
>>> Subject: [Moses-support] Multi-threading / Boost lib / compile error for 
>>> threaded Moses
>>> To: "Kenneth Heafield"<[email protected]>
>>> Cc:<[email protected]>
>>>
>>>
>>> this is the last thing i will post here on this subject:
>>>
>>> debugging with a single thread running invokes the threading code.
>>> ***if you suspect that this is somehow broken, then you need to debug
>>> without it***.  it is that simple.
>>>
>>> running gdb in single thread mode still uses threading.
>>>
>>> Miles
>>>
>>> On 22 September 2011 11:28, Kenneth Heafield<[email protected]>  wrote:
>>>> But I don't see a use case for it.  I can run gdb just fine on a
>>>> multithreaded program that happens to be running one thread.  And the
>>>> stderr output will be in order.
>>>>
>>>> On 09/22/11 11:21, Miles Osborne wrote:
>>>>> should someone want to debug with no threading, then there would need
>>>>> to be a mess of ifdefs removing all support for threading.  i agree,
>>>>> this will be a pain to deal with, but this is what debugging with no
>>>>> threads means.
>>>>
>>>
>>>
>>> --
>>> The University of Edinburgh is a charitable body, registered in
>>> Scotland, with registration number SC005336.
>>>
>>> _______________________________________________
>>> Moses-support mailing list
>>> [email protected]
>>> http://mailman.mit.edu/mailman/listinfo/moses-support
>>>
>>>
>>>
>>> _______________________________________________
>>> Moses-support mailing list
>>> [email protected]
>>> http://mailman.mit.edu/mailman/listinfo/moses-support
>>
>> _______________________________________________
>> Moses-support mailing list
>> [email protected]
>> http://mailman.mit.edu/mailman/listinfo/moses-support
>>
>
>
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support

Reply via email to