Re: [Axiom-commit] [Axiom-developer] Re: SF.net SVN: axiom: [266]branches/build-improvements/src

[Gabriel Dos Reis]

"Bill Page" <address@hidden> writes:

| On November 17, 2006 9:58 PM Gaby wrote:
| > ...
| > | 
| > | What components in particular were not built?
| > 
| > 
| > graph, sman, hyper and input.
| >
| 
| These all built for me with no visible problems.

Yes.  See the message I just sent.

| > | What error message should I look for in the log?
| > 
| > There is no dirct error message on console.  However, when I looked
| > at build/$target/trace, there are many instances of hyperdoc
| > complaining it does not understand \spad.
| >
| 
| I do not see any such messages from hyperdoc in my
| 
|   build/i686-pc-linux/trace

grep for "Unexpected HT command:"

[...]

| > | What machine were you using for the test?
| > 
| > I did the binary search on my laptop, i686-pc-linux-gnu.
| > I'm cross checking on a x86-suse-linux.
| >
| 
| axiom-developer is seen as i686-pc-linux. I would not expect
| results to differ significantly from yours.

Yes.  Now, I think I understand what is going go.  See the other
message.  AXIOMsys is built twice.  The first time it is tested, the
second time it is not -- and that is no good.

First, we should not build twice.  Second, if we must build twice,
then we must test the second version -- the one that is installed.

Finally, I think we must rewrite the patch to src/etc/ that introduces
this new behaviour.



Th reason why I thought the branch is broken is that I have a script
that "watches" the output of the build log and it got confused by
the new behaviour.  Then I spend the whole day on it :-/

-- Gaby