[ADMB Users] problem with gcc on arm processor

Johnoel Ancheta johnoel at hawaii.edu
Tue Dec 2 12:13:22 PST 2014


Thanks Dave,

Committed the changes

http://www.admb-project.org/redmine/projects/issues/repository/revisions/2740/diff/

On Fri, Nov 28, 2014 at 1:25 PM, dave fournier <davef at otter-rsch.com> wrote:

>  On 11/28/2014 03:14 PM, Johnoel Ancheta wrote:
>
> I posted the change.
>
> explicity oine gets rid of
>
>
>
> struct {
>     integer mp, lp;
>     doublereal gtol, stpmin, stpmax;
> } lb3_1 = { .mp = 6, .lp = 6, .gtol = .9, .stpmin = 1e-20, .stpmax = 1e20};
>
> and replaces it with the two step
>
> struct kkkludge{
>     integer mp, lp;
>     doublereal gtol, stpmin, stpmax;
> };
> struct kkkludge lb3_1 = {  6, 6, 0.9, 1e-20, 1e20};
>
> where you first define the kkludge struct and then define an instance and
> initialize it without the dot stuff.
>
>
>
>
>  The intel also complains.  Let me know if you have a solution.
>
> On Friday, November 28, 2014, dave fournier <davef at otter-rsch.com> wrote:
>
>> I built admb on a quad core arm processor in a TV dongle.  Wanted to
>> compare the power to my old  486.
>>
>> The 486 took 38 seconds to run the catage example. The TV dongle runs it
>> in less than a second. Neat.
>>
>> However the version of gcc  chokes on the code in lbfgs.cpp. Around line
>> 250 it seems to prefer a more pedestrian way of
>> declaring and initializing a struct.
>>
>> /* Common Block Declarations */
>>
>> #if !defined(__SUNPRO_CC) && !defined(__OPEN64__) && !defined(_MSC_VER)
>>
>> struct kkkludge{
>>     integer mp, lp;
>>     doublereal gtol, stpmin, stpmax;
>> };
>> struct kkkludge lb3_1 = {  6, 6, 0.9, 1e-20, 1e20};
>>
>> // comment out
>> /*
>> struct kkkludge lb3_1 = {  6, 6, 0.9, 1e-20, 1e20};
>> */
>>
>> there appears to be a problem with "."
>>
>> Changin the line
>>
>> struct kkkludge lb3_1 = {  6, 6, 0.9, 1e-20, 1e20};
>>
>> to
>>
>> struct kkkludge lb3_1 = {  6, 6, .9, 1e-20, 1e20};
>>
>> produces an error.
>> _______________________________________________
>> Users mailing list
>> Users at admb-project.org
>> http://lists.admb-project.org/mailman/listinfo/users
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.admb-project.org/pipermail/users/attachments/20141202/837be256/attachment.html>


More information about the Users mailing list