Project

General

Profile

Design #932

CoCoALib configuration: BOOST dependency

Added by John Abbott over 7 years ago. Updated about 4 years ago.

Status:
Closed
Priority:
Normal
Assignee:
John Abbott
Category:
Portability
Target version:
CoCoALib-0.99700
Start date:
30 Sep 2016
Due date:
% Done:

100%

Estimated time:
1.66 h
Spent time:
1.60 h

Description

Currently we have a single configuration process (script) for both CoCoALib and CoCoA-5. This is not ideal!

Gereon was surprised that a CoCoALib program he had written needed to link in BOOST code before it could run. I, too, was surprised when he told me (since I believed CoCoALib to be BOOST-free).

What happens is that the configure script looks for BOOST automatically; and, if found, takes note of where it is. If BOOST is present then inside CoCoALib the ThreadsafeCounter code will switch to using BOOST code (even if CoCoA_THREADSAFE_HACK is not set).

What to do? Discuss!

Related issues

Related to CoCoALib - Feature #319: BOOST -- how it could help in CoCoALibClosed2013-02-21

Related to CoCoALib - Design #933: Separate configure scripts for CoCoALib and CoCoA-5In Progress2016-09-30

History

#1 Updated by John Abbott over 7 years ago

  • Related to Feature #319: BOOST -- how it could help in CoCoALib added

#2 Updated by John Abbott over 7 years ago

As far as I can see, the code in ThreadsafeCounter will probably become much simpler when we switch to C++11 (since atomic types are part of the language).

ThreadsafeCounter is used for creating anonymous symbols (see symbol.H) and ring IDs (see ring.C).

#3 Updated by John Abbott over 7 years ago

A reasonable solution might simply to be to remove the BOOST dependent code in ThreadsafeCounter. This would mean that threadsafety is not possible with a pre-C++11 compiler (though you would probably have to create very rapidly many anonymous symbols or many rings to trigger a problem).

#4 Updated by John Abbott over 7 years ago

  • Related to Design #933: Separate configure scripts for CoCoALib and CoCoA-5 added

#5 Updated by John Abbott over 7 years ago

  • Status changed from New to In Progress
  • % Done changed from 0 to 10

The CoCoA configure script complains if someone specifies "with Normaliz" but "without BOOST". The code in CoCoALib does not depend on BOOST at all.

Bruns reported by email (2016-11-29) some problems with CoCoALib wanting to use BOOST (perhaps as an attempt to make Normaliz and cocoa5(?) communicate with Sage).

#6 Updated by John Abbott over 7 years ago

I have added a --no-boost option to the configure script; this disables automatic searching for the BOOST headers and libraries.

configure used to give an error if --with-normaliz was specified and BOOST was absent; however, a quick check showed that there is no true dependency: i.e. disabling the check in the configure script, and then compiling and running everything worked fine.

I do note that nm finds some BOOST symbols in the executables involving Normaliz, but they have probably come from just the BOOST header files... so absence of BOOST is not a problem.

I shall comment out the check in configure for BOOST being required if --with-normaliz has been specified. I prefer not to delete it yet in case there is some nasty surprise waiting for me... :-/ [fine after 2 years: deleted the cruft 2019-03-25]

#7 Updated by John Abbott about 4 years ago

  • Status changed from In Progress to Closed
  • Assignee set to John Abbott
  • Target version changed from CoCoALib-1.0 to CoCoALib-0.99700
  • % Done changed from 10 to 100
  • Estimated time set to 1.66 h

Since we moved to C++11/C++14 we can use an atomic C++ counter instead of the threadsafe code; so I have already deleted the threadsafecounter code (some time ago).

This issue is now completely resolved (I hope!)

Also available in: Atom PDF