[erlang-questions] Problems cross-compiling OTP on ARM

Bjorn Gustavsson bgustavsson@REDACTED
Sat Sep 27 08:15:49 CEST 2008


On Fri, Sep 26, 2008 at 7:15 PM, Mikael Pettersson <mikpe@REDACTED> wrote:

> Matthias Lang writes:
>  > On Friday, September 26, Corrado Santoro wrote:
>  >
>  > > The strange thing is that I've disabled hipe (./configure
>  > > --disable_hipe), but it seems that hipe is still considered?
>  >
>  > You've pretty much figured it out: --disable_hipe would more accurately
>  > be
> --disable_some_parts_of_hipe_but_not_others_which_is_fine_unless_you're_cross_compiling_in_which_case_its_a_bit_of_a_pain
>
> The reason for that is that dialyzer for reasons I don't know/remember
> uses parts of HiPE, and it was decided (jointly by the dialyzer group
> and OTP) to keep parts of HiPE even if --disable-hipe is in effect so
> that dialyzer can work.
>

Yes, that is correct.

Actually, I think that the current version of Dialyzer does not use HiPE as
much as it used to.
It used to run the first pass of HiPE compiler and do its analysis on the
icode format, but as
far as I know, Dialyzer now does the entire analysis using Core Erlang and
only uses a few
modules in hipe/cerl (which are pure Erlang, and not dependent on any HiPE
support in the
run-time system).

Tobias or Kostis could confirm those details, but I think that it would be
enough to include
only the hipe/cerl sub-directory to support Dialyzer.

/Bjorn

-- 
Björn Gustavsson, Erlang/OTP, Ericsson AB
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20080927/80a10afc/attachment.htm>


More information about the erlang-questions mailing list