move src/pbc_info.c to examples/c/pbc_sample.c

[rurban]

r3904 added src/pbc_info.c sample program

This should be moved to examples/c or examples/tools.

It is no info program, and it *changes* the pbc's by adding some sample sections. The info program is called pbc_dump.

[jkeenan]

Replying to rurban:

r3904 added src/pbc_info.c sample program

I looked at this TT for the first time today. I am no bytecode expert, so some of the following comments/questions may be naive. But, anyhow ...

The source code file is src/pbc_info.c. It is compiled into the stand-alone executable pbc_info in the very last step of make.

src/pbc_info.c
cc -o pbc_info \
    src/pbc_info.o \
    -Wl,-rpath=/home/jimk/work/parrot/blib/lib -L/home/jimk/work/parrot/blib/lib -lparrot -lpthread -lm -L/usr/lib  -licuuc -licudata -lpthread -lm -lnsl -ldl -lm -lcrypt -lutil -lpthread -lgmp -lreadline  -L/usr/local/lib -Wl,-E  

This should be moved to examples/c or examples/tools. It is no info program, and it *changes* the pbc's by adding some sample sections. The info program is called pbc_dump.

I can't say whether pbc_info changes any PBCs; I saw no evidence of that.

But Reini is correct in reporting that it is not very informative, particularly in comparison with pbc_dump. I will attach 10 files which display the output of ./pbc_info and ./pbc_dump on each of 5 .pbc files as run on Linux i386. Except where they bytecode wasn't appropriate for Linux -- where both programs failed -- the information supplied by pbc_info was vastly inferior to that supplied by pbc_dump.

If that is the case, then I have to ask why we should have 'make' build the 'pbc_info' at all. Let's cut it out entirely. And if we do that, we shouldn't leave src/pbc_info.c where it is. But if we move it to somewhere under examples/ as Reini suggested, then we first should ask, 'What exactly is it an example of?'

Thoughts? Thank you very much.

kid51

[jkeenan]

Replying to jkeenan:

If that is the case, then I have to ask why we should have 'make' build the 'pbc_info' at all. Let's cut it out entirely.

No one objected to this -- but, then again, nobody said anything at all. So I now formally propose the deprecation of pbc_info as a Parrot feature. I have begun work on this in the pbc_info_remove branch. Attached is a patch which represents the work there. The following make targets ran or tested satisfactorily: all, test, buildtools_tests, manifest_tests and most of codetest. I can't quite figure out why examples/c/pbc_info.c is now failing the c_function_docs coding standard; other eyeballs welcome.

I propose that we apply this patch to trunk at the post-1.5 deprecation point, i.e., after the third Tuesday in July.

I grepped the repository for pbc_info and PINFO and removed whatever I found that wasn't tied down, with one exception: I did not touch the Cygwin ports files, as they were developed in patch form. I did, however, modify all the other ports files where those strings were found.

And if we do that, we shouldn't leave src/pbc_info.c where it is. But if we move it to somewhere under examples/ as Reini suggested, then we first should ask, 'What exactly is it an example of?'

For lack of any better idea, I have moved pbc_info.c to examples/c/.

Comments?

Thank you very much.

kid51

[jkeenan]

diff between trunk and pbc_info_remove branch

[jkeenan]

Replying to jkeenan:

For lack of any better idea, I have moved pbc_info.c to examples/c/.

Merged pbc_info_remove branch into trunk at r40204. All tests pass. Closing ticket.