[haiku-commits] Re: buildbot failure in HAIKU on (whole buildset)

  • From: Jessica Hamilton <jessica.l.hamilton@xxxxxxxxx>
  • To: haiku-commits@xxxxxxxxxxxxx
  • Date: Mon, 15 Jun 2015 17:07:29 +1200


On 15/06/2015, at 5:01 pm, Adrien Destugues <pulkomandy@xxxxxxxxxxxxx> wrote:


The grist for nsparser.h in the first line looks wrong (it's missing the
x86_gcc2). However I'm not sure why and how to fix that.
I have just added an explicit dependency in hrev49297 to address the issue.

Thanks, but I would prefer a generic solution fixing this in the Lex or Yacc
rule, or maybe moving the invocation of these rules outside the architecture
specific part (the generated file is exactly the same, so why is it tagged
architecture specific in the first place?)

I had tried moving it outside the architecture specific part, but couldn’t get
Jam to do the right thing =/ And yes, I’d prefer a generic fix too...

In terms of reproducing, it's fairly straightforward to do:
find objects -name 'nslexer.*' | xargs rm
find objects -name 'nsparser.*' | xargs rm
jam -q '<src!kits!network!netresolv!x86_gcc2>nslexer.o'
Then go crazy poking the jam files :p

I tried this several times on different machines, but here the build goes
just fine, even with jam -j16 or -j32. I'm building from Haiku however, maybe
this somehow avoids the problem.

You shouldn’t be passing a -jN flag for reproducing, although would be
surprised to see the exact steps above succeed without the fix. I’m also
building on Haiku ;-)

Other related posts: