On Wed, May 25, 2011 at 8:25 PM, Jon Rosebaugh <span dir="ltr"><<a href="mailto:chairos@gmail.com">chairos@gmail.com</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
it should be possible to have a script<br>
cross-compile those programs and set up a little shell script to run<br>
them and output the cross_config file. Then you would just copy that<br>
to your target system, run it, and there you go. Any reason this<br>
hasn't been done yet?<br></blockquote><div><br>Two: lack of manpower and also lack of portability. How to run them? This varies a lot from machine to machine and many cross compilation environments do not even provide a way to do that. Furthermore, many of the machine tests are automatically generated by autoconf and extracting them into separate programs requires some additional effort, which is a bit redundant.<br>
</div><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
In addition, there seems to be a problem with configure around line<br>
9038; normally the test programs are surrounded by a check that skips<br>
them if the cross_config has a definition, but the test for<br>
HAVE___BUILTIN_RETURN_ADDRESS seems to miss this, so that a<br>
cross-compilation will always fail.<br></blockquote></div><br clear="all">Thanks for reporting. For such a clear case in which you have identified the problem so perfectly it is generally more useful to submit a bug report -- I am very disorganized and tend to forget emails, while the SF bug tracker is always there to remind me how lousy a maintainer I am :-)<br>
<br>Thanks again for your interest and your help.<br><br>Juanjo<br><br>-- <br>Instituto de Física Fundamental, CSIC<br>c/ Serrano, 113b, Madrid 28006 (Spain) <br><a href="http://juanjose.garciaripoll.googlepages.com" target="_blank">http://juanjose.garciaripoll.googlepages.com</a><br>