bug-bison
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Bison 3.4.1 released [stable]


From: Akim Demaille
Subject: Re: Bison 3.4.1 released [stable]
Date: Sun, 26 May 2019 15:57:58 +0200

Hi Dennis!

Thanks a lot for your efforts!

> Le 25 mai 2019 à 19:11, Dennis Clarke <address@hidden> a écrit :
> 
> Here is the crap from Oracle :
> 
> corv $
> corv $ /bin/bash --version
> GNU bash, version 3.2.57(1)-release (sparc-sun-solaris2.10)
> Copyright (C) 2007 Free Software Foundation, Inc.
> corv $

It's not recent, indeed, but it should work for the test suite.

>> Something like make check SHELL=/bin/bash should work.
> 
> I agree totally.
> 
> Should.
> 
> corv $
> corv $ /bin/bash --version
> GNU bash, version 3.2.57(1)-release (sparc-sun-solaris2.10)
> Copyright (C) 2007 Free Software Foundation, Inc.
> corv $ SHELL=/bin/bash ; export SHELL
> corv $
> 
> .
> .
> .
> 
> FAIL: examples/c/mfcalc/mfcalc.test

Well, this appears to be caused by my crap, not Oracle's.  I'm sorry to report 
that I believed that the test suite was using SHELL, but it didn't (back to 
"should")...

I apologize for the waste of time that followed (including installing a more 
modern version of Bash).

If you are still willing to try Bison 3.4 on these boxes, please give a try to 
this version, which includes the patch at the bottom of this message.

https://www.lrde.epita.fr/~akim/private/bison/bison-3.4.1.5-1dcd6.tar.gz
https://www.lrde.epita.fr/~akim/private/bison/bison-3.4.1.5-1dcd6.tar.xz

>> Well, the thing is that it's a pain to be compliant with them, if it's 
>> useless.
> 
> Pain?  Let me tell you about pain.  I had a pile of Oracle/Solaris zones
> running in production within a finance corp and had to get up to date
> Apache httpd there as well as OpenSSL with latest TLS 1.3 and then php
> also.  A horror show nightmare. A nightmare I tell you.

Wow...  I had no idea you faced such challenges, but I can imagine
(well, probably only vaguely) how tricky it must be to shoehorn
new pieces of software onto an old platform.

> Simple solution?  Install a deck load of Debian and FreeBSD boxen. All
> problems go away with 10Gbit LACP and backend storage and VMware ESXi
> and other modern solutions.

Yes, I can imagine how suddenly friction and viscosity is reduced.

> Anyways Solaris on sparc is a curse.

It used to be a beautiful environment.  Their chips are so much better
than Intel's (well, Itanium was promising, but actually it's not even
Intel that started it).

But now all this is bit-rotten.

> I think I am qualified to say so.
> It has three features : very expensive, very slow, very power hungry and
> runs hot as anything you can imagine.  Four.  Four features.

:) :) :)

> Also nothing else runs on sparc.  Not any recent sparc.
> 
> Anyways, rant aside, there are shells out there other than bash and so
> it may be of some limited value to make the scripts and such really
> baseline old school compliant.  However is it worth the effort?

Well, it really depends on the cost.  Migrating to full support for
the native shell is probably too painful (and with a negative impact
for new users: the "example/test" shell script is installed with
the examples and meant to be read by end users, whom I'd like to
protect from antique idioms such as `...` (which cannot be nested
safely) instead of $(...)).  But supporting properly $SHELL is
definitely Bison should do.

> Here we flip a coin.  Personally I choose to not ruin a nice Saturday.

I hope you enjoyed the rest of it!

Thanks again, Dennis, for having spent so much time on Bison's test suite.

Cheers!


commit 1dcd6068dd3cbd5fd8e041fd6351651b0bdb84ef
Author: Akim Demaille <address@hidden>
Date:   Sun May 26 15:29:37 2019 +0200

    tests: take SHELL into account
    
    Reported by Dennis Clarke.
    http://lists.gnu.org/archive/html/bug-bison/2019-05/msg00053.html
    
    * examples/local.mk, tests/local.mk: here.

diff --git a/examples/local.mk b/examples/local.mk
index ff04345c..db6fe650 100644
--- a/examples/local.mk
+++ b/examples/local.mk
@@ -33,7 +33,7 @@
 ## which guarantees that parse.y is indeed shipped.
 
 dist_noinst_SCRIPTS = %D%/extexi %D%/test
-TEST_LOG_COMPILER = $(top_srcdir)/%D%/test
+TEST_LOG_COMPILER = $(SHELL) $(top_srcdir)/%D%/test
 
 AM_CXXFLAGS =                                                  \
   $(WARN_CXXFLAGS) $(WARN_CXXFLAGS_TEST) $(WERROR_CXXFLAGS)
diff --git a/tests/local.mk b/tests/local.mk
index c8ee3f02..7c233042 100644
--- a/tests/local.mk
+++ b/tests/local.mk
@@ -85,7 +85,7 @@ $(TESTSUITE): $(TESTSUITE_AT)
 ## -------------------- ##
 
 # Move into %D%/ so that testsuite.dir etc. be created there.
-RUN_TESTSUITE = $(TESTSUITE) -C %D% $(TESTSUITEFLAGS)
+RUN_TESTSUITE = $(SHELL) $(TESTSUITE) -C %D% $(TESTSUITEFLAGS)
 check_SCRIPTS += $(BISON) %D%/atconfig %D%/atlocal
 RUN_TESTSUITE_deps = all $(TESTSUITE) $(check_SCRIPTS)
 




reply via email to

[Prev in Thread] Current Thread [Next in Thread]