Make bzip2 a standard (instead of base) package

[jdemeyer]

Now that we have support for gzipped Sage packages (#12602), we can easily make bzip2 a standard package instead of a base package. This will further the goal of decreasing the number of base packages (ideally to zero), see also #12631.

spkg: http://boxen.math.washington.edu/home/jdemeyer/spkg/bzip2-1.0.6.spkg (bzip2 compressed, to be converted to gzip when merged)

apply attachment: 12102_bzip2_standard.patch to SAGE_ROOT.

delete the untracked file spkg/base/bzip2-1.0.5.tar.gz.

Depends on #12479
Depends on #10492
Depends on #12602

Component: packages: standard

Author: Jeroen Demeyer

Reviewer: John Palmieri

Merged: sage-5.0.beta11

Issue created by migration from https://trac.sagemath.org/ticket/12102

[jdemeyer]

Dependencies: #12602

[jdemeyer]

Author: Jeroen Demeyer

[jdemeyer]

Changed dependencies from #12602 to #12479, #10492, #12602

[jdemeyer]

comment:6

Attachment: 12102_bzip2_standard.patch.gz

[jhpalmieri]

comment:7

Is this ready for review?

[jdemeyer]

comment:8

If you want, it's ready for review. But to be honest, I haven't really tested it.

[jhpalmieri]

comment:10

I see that the old package compiles with commands like this:

gcc -fPIC -c blocksort.c

while the new one compiles with

gcc -fPIC -O2 -g  -c blocksort.c

-O2 seems like a fine idea, but do we need -g? I don't know much about debugging: since this is a pretty well-established package, there aren't like to be bugs in it, so can we omit -g, or would that mess up debugging for other parts of Sage? Should we add a test for SAGE_DEBUG so that we at least have the option to turn it off? I notice that the new libbz2.a is more than twice the size of the old one, at least on sage.math.

[jdemeyer]

comment:11

Replying to @jhpalmieri:

-O2 seems like a fine idea, but do we need -g?

I thought it was "standard practice" to compile all Sage packages using -g (regardless of SAGE_DEBUG). With GCC, files only get larger when compiled with debug information, but the code is exactly the same. So programs should run equally fast with or without debugging information (except for the I/O effects of having larger files).

[jhpalmieri]

Reviewer: John Palmieri

[jhpalmieri]

comment:12

It looks to me as though SAGE_DEBUG is used somewhat inconsistently. Since William (among others) is concerned about the size of the Sage distribution, I think at some point we should consider not using -g. Maybe don't use it for bdists? Anyway, this is not something that should be decided here, and I'm happy with the changes in this ticket.

[jdemeyer]

comment:13

Let me mention that I also tested upgrading from sage-4.5 and sage-4.8, no problems.

[jdemeyer]

Merged: sage-5.0.beta11

[jhpalmieri]

comment:16

By the way, the file spkg/base/README.txt should be modified to reflect this change. Is the directory spkg/base going to disappear altogether soon? If so, we don't have to worry about this.

[jdemeyer]

comment:17

Replying to @jhpalmieri:

By the way, the file spkg/base/README.txt should be modified to reflect this change.

Going to do this in #12631.

Is the directory spkg/base going to disappear altogether soon?

Probably not. Eventually, I would like prereq to become a top-level ./configure for Sage, but I haven't thought nearly enough about this to see how that would work.