Re: summary data about all packages in the repository

[Aleksey Cheusov <cheusov%tut.by@localhost>]

If anyone interested, I'm working on this.

I think I have some interesting intermediate results.

wip/pkg_cmp_summary package:
  - pkg_cmp_summary script compares two summaries, generated
    by pkg_info -X, wip/pkg_src_summary or anything else.

   Using this simple tool it is easy to check your installed
   packages against binary repository or pkgsrc source tree.
   See below.

wip/pkg_src_summary package:
  - pkg_src_summary generates summary in pkg_info -X format
    (exactly what I want) but slowly.
  - pkg_src_micro_summary also generates summary but produces PKGNAME
    and PKGPATH fields only. But it does this approximately 10(!!!) times
    faster than pkg_src_summary.

    Using micro summary it is possible to
      a) check your installed
         packages against pkgsrc tree _drammatically_ faster than
         pkg_chk -nus does .

         Example (451 installed packages, Athlon-800, NetBSD):

         0 tmp>time pkg_info -X '*' > /tmp/inst.txt 
         0.73s real     0.57s user     0.13s system
         0 tmp>time sed -n 's/^PKGPATH=//p' inst.txt | pkg_src_micro_summary  > 
pkgsrc.txt                   
         cd: can't cd to x11/Xrandr
         Bad package x11/Xrandr, skipped
         cd: can't cd to wip/netbsd-shlock
         Bad package wip/netbsd-shlock, skipped
         36.81s real    29.94s user    11.81s system
         0 tmp>time pkg_cmp_summary inst.txt pkgsrc.txt | grep '^[^=]' > 
cmp_result.txt                      
         0.10s real     0.08s user     0.01s system
         0 tmp>head -50 cmp_result.txt
         < arts 1.5.7 1.5.8
         < nas 1.9 1.9nb1
         < emacs 22.1nb1 22.1nb2
         < libuxre 070227 070715
         < nano 1.2.4nb1 2.0.7
         < pcre 7.3 7.4nb1
         < perl 5.8.8nb5 5.8.8nb6
         < glib2 2.14.1 2.14.4
         < fontconfig 2.4.2nb3 2.5.0
         < png 1.2.20 1.2.23
         < gnuplot 4.0.0nb7 4.2.2nb1
         < gawk 3.1.5nb1 3.1.6
         < fvwm 2.4.20 2.4.20nb1
         ? libtheora 1.0alpha7 1.0beta2
         ...
         0 tmp>

      b) what packages in binary repository needs to be upgraded
         (based on PKGNAME and PKGREVISION variables from Makefile)
      c) ?

Any thoughts, suggestions?

> Does anybody know a program/way to obtain a summary information
> about ALL packages in pkgsrc/ repository?

> Something like 'make print-summary-data' or any other textual format
> (foprmat of INDEX file is also good)

>   example (make print-summary-data):

>     depends archivers/gzip
>     build_depends archivers/gzip
>     conflicts archivers/gzip gzip-base-[0-9]* gzip-info-[0-9]*
>     index archivers/gzip gzip-1.3.12nb1
>     htmlname archivers/gzip <a 
> href="../../archivers/gzip/README.html">gzip-1.3.12nb1</a>
>     homepage archivers/gzip http://www.gnu.org/software/gzip/gzip.html
>     wildcard archivers/gzip gzip-[0-9]*
>     comment archivers/gzip Compress or expand files
>     license archivers/gzip
>     onlyfor archivers/gzip any
>     notfor archivers/gzip any
>     maintainer archivers/gzip pkgsrc-users%NetBSD.org@localhost
>     categories archivers/gzip archivers
>     descr archivers/gzip archivers/gzip/DESCR
>     prefix archivers/gzip /usr/pkg

> but for ALL packages.

> It's easy to scan all packages and run 'make print-summary-data'
> for each package but this takes looooooong time.
> I need a fast (more or less) solution. Incremental approach is good.
> I hope somebody already did this task.

> -- 
> Best regards, Aleksey Cheusov.


-- 
Best regards, Aleksey Cheusov.