compiler problems with libadacrypt-dev-0.1.3

compiler problems with libadacrypt-dev-0.1.3

[batoubrioche]

For a project I need to use some of the libadacrypt library's facilities. I downloaded the most recent version of libadacrypt-dev-0.1.3 here at sourceforge

http://sourceforge.net/projects/libadacrypt-dev/

First of all, I programmed some small test units with the library's big number support.

But I get these error messages while compiling the crypto-types.ads source:

Compiling ...
crypto-types.ads:108:34: first argument for shift must have size 8. 16, 32 or 64
crypto-types.ads:109:34: first argument for shift must have size 8. 16, 32 or 64
crypto-types.ads:291:30: incorrect intrinsic subprogram, see spec
crypto-types.ads:292:30: incorrect intrinsic subprogram, see spec
Done--error detected.

I have no clue what is wrong here. Could someone tell me what is wrong with this source from the libadacrypt library?

Btw. I use the gnat-gpl-2013-i686-pc-mingw32 package from Adacore for Windows.

Thank you very much for your help.

Batou

Re: compiler problems with libadacrypt-dev-0.1.3

[Simon Wright]

batoubrioche@googlemail.com writes:

> For a project I need to use some of the libadacrypt library's
> facilities. I downloaded the most recent version of
> libadacrypt-dev-0.1.3 here at sourceforge
>
> http://sourceforge.net/projects/libadacrypt-dev/
>
> First of all, I programmed some small test units with the library's
> big number support.
>
> But I get these error messages while compiling the crypto-types.ads source:
>
> Compiling ...
> crypto-types.ads:108:34: first argument for shift must have size 8. 16, 32 or 64
> crypto-types.ads:109:34: first argument for shift must have size 8. 16, 32 or 64
> crypto-types.ads:291:30: incorrect intrinsic subprogram, see spec
> crypto-types.ads:292:30: incorrect intrinsic subprogram, see spec
> Done--error detected.
>
> I have no clue what is wrong here. Could someone tell me what is wrong
> with this source from the libadacrypt library?

I've tried this with GCC 4.8.1 ... and get the same.

The problem is that, for the first error, the code says

   function Shift_Left  (Value : Natural; Amount : Natural) return Natural;

and Natural'Size (on this machine, x86_64-apple-darwin12) is 31, not
32. Likewise for the second error.

Having "corrected" this by using Integer for Value and the result, the
next problem is with crypto-symmetric-algorithm-tripledes.ads:103 to
:106; this is less tractable, since the code is trying to implement
intrinsic shifts for a 6-bit type. It is in a section of code marked
obsolete, though!

I posted bugs to https://sourceforge.net/p/libadacrypt-dev/bugs/.

Re: compiler problems with libadacrypt-dev-0.1.3

[batoubrioche]

Dear Simon

Am Sonntag, 30. März 2014 16:43:52 UTC+2 schrieb Simon Wright:
> I posted bugs to https://sourceforge.net/p/libadacrypt-dev/bugs/.

Thank you very much for your help. I think someone need to be a real expert to track down all those problems within this code.

I hope, someone still takes care of this library and will correct all those incompatibility errors.

kind regards,
Batou

Re: compiler problems with libadacrypt-dev-0.1.3

[Simon Wright]

batoubrioche@googlemail.com writes:

> Am Sonntag, 30. März 2014 16:43:52 UTC+2 schrieb Simon Wright:
>> I posted bugs to https://sourceforge.net/p/libadacrypt-dev/bugs/.
>
> Thank you very much for your help. I think someone need to be a real
> expert to track down all those problems within this code.
>
> I hope, someone still takes care of this library and will correct all
> those incompatibility errors.

A further bug at https://sourceforge.net/p/libadacrypt-dev/bugs/4/ - I'm
now stuck, you might have better luck on a 32-bit machine.

(I "fixed" the Bit6 problem by implementing the functions to raise
Program_Error - might be OK since in a section marked Obsolete!)

Re: compiler problems with libadacrypt-dev-0.1.3

[Christian Forler]

On 30.03.2014 14:54, batoubrioche@googlemail.com wrote:
> For a project I need to use some of the libadacrypt library's facilities. I downloaded the most recent version of libadacrypt-dev-0.1.3 here at sourceforge
> 
> http://sourceforge.net/projects/libadacrypt-dev/
> 
> First of all, I programmed some small test units with the library's big number support.
> 
> But I get these error messages while compiling the crypto-types.ads source:
> 
> Compiling ...
> crypto-types.ads:108:34: first argument for shift must have size 8. 16, 32 or 64
> crypto-types.ads:109:34: first argument for shift must have size 8. 16, 32 or 64
> crypto-types.ads:291:30: incorrect intrinsic subprogram, see spec
> crypto-types.ads:292:30: incorrect intrinsic subprogram, see spec
> Done--error detected.
> 
> I have no clue what is wrong here. Could someone tell me what is wrong with this source from the libadacrypt library?
> 
> Btw. I use the gnat-gpl-2013-i686-pc-mingw32 package from Adacore for Windows.
> 

The library at sourceforge is obsolete.

Please use the current version from github

https://github.com/cforler/Ada-Crypto-Library

BTW. I'm the maintainer of this library.


Best regards,
Christian

Re: compiler problems with libadacrypt-dev-0.1.3

[Simon Wright]

Christian Forler <christian.forler@uni-weimar.de> writes:

> The library at sourceforge is obsolete.

Could we ask you, then, to please update the SF site to say so?

Thanks.

Re: compiler problems with libadacrypt-dev-0.1.3

[Christian Forler]

On 02.04.2014 14:53, Simon Wright wrote:
> Christian Forler <christian.forler@uni-weimar.de> writes:
> 
>> The library at sourceforge is obsolete.
> 
> Could we ask you, then, to please update the SF site to say so?

I would love to update the FS version but there is a little problem.

I locked myself out ( I forget my password + I changed my email
address). I tried several times to convince SF to reset my password, but
they have ignored me, so far. :-(


Best regards,
Christian

Re: compiler problems with libadacrypt-dev-0.1.3

[batoubrioche]

Am Mittwoch, 2. April 2014 10:54:17 UTC+2 schrieb Christian Forler:
> The library at sourceforge is obsolete.
> 
> 
> 
> Please use the current version from github
> 
> 
> 
> https://github.com/cforler/Ada-Crypto-Library

Thank you very much Christian. I tried this newer version and now everything works well.

Are there still any bugs I should be aware of?

Best regards,

Batou

Re: compiler problems with libadacrypt-dev-0.1.3

[Christian Forler]

On 08.04.2014 17:57, batoubrioche@googlemail.com wrote:
> Am Mittwoch, 2. April 2014 10:54:17 UTC+2 schrieb Christian Forler:
>> The library at sourceforge is obsolete.
>>
>>
>>
>> Please use the current version from github
>>
>>
>>
>> https://github.com/cforler/Ada-Crypto-Library
> 
> Thank you very much Christian. I tried this newer version and now everything works well.
> 
> Are there still any bugs I should be aware of?


The defult random source of the build in random number generator is
"/dev/urandom". If you work under Windows you have to add a new random
soure.


Best regards,
Christian