[PC-BSD Pbi-dev] Pbi-dev Digest, Vol 8, Issue 1

Terry Poulin bigboss64 at ippimail.com
Sat Aug 18 14:34:14 PDT 2007


> Date: Fri, 17 Aug 2007 23:35:03 -0300
> From: "Charles A. Landemaine" <charles at pcbsd.org>
> Subject: [PC-BSD Pbi-dev] News on the PBI development front!
> To: "pbi-dev at lists.pcbsd.org" <pbi-dev at lists.pcbsd.org>
> Message-ID: <op.tw8hgpbc0bhqog at sony>
> Content-Type: text/plain; format=flowed; delsp=yes;
> 	charset=iso-8859-15
>
> I had an interesting chat today with Alex (mal.exe) about PBIs, and in our
> continuous quest for better quality, I developed a simple tool that makes
> life easier for PBI testers to report their test of PBIs. PBI testers only
> have to fill out a simple form, generate the report and paste it into the
> forum thread :)
>
> Give it a whirl: http://test.pbidir.com
>

You might consider having it E-Mail a copy of the report to the relevant
people (like Dedicated PBI Testers, person(s) uploading to pbidir, and the
Creator of the PBI) automatically. It's just an idea,


Over on the site where I call home. We have several report forms. It
allows us to provide an easy way for people to report to us any issues and
do common paper work for their roles on the team. And have a message sent
to the people that should be informed, automatically for them ;-)


> Also, Gon, Terry and Alex among others, will be pleased to know that it is
> now *OFFICIAL*, PBIs that use pkg_add don't follow the quality guidelines
> anymore: http://faqs.pcbsd.org/18_296_en.html
>

Oh crap, that will play havoc with www.pbidir.com .... I hope the real
player codecs I have for xine/mplayer work dang good if any thing happens
to the linux-realplayer port ;-)


Note to self, delete all of the half finished PBI Docs I have in my home
directory...


> The points discussed by Terry, Renato and Alex on the forum are
> insightful, and we have tried to make PBIs more stable with several
> measures. The PBI Updater now works with an increasing number of
> applications also.
>
> Here are some suggestions to increase quality of PBIs:
>   - Each PBI should be tested and approved by 3 different people
>   - Tests should be done on fresh PC-BSD installations
>   - Each tester should check all points of the requirements (ideally with
> the report generator!)
>
> Could we work this way, guys? Any other idea?
>


I don't even remember the last time I logged into the forum....  I should
try to drop in after work and see what I've missed. I think I'm off Friday
after next.


This is what _I_ personally have been thinking of doing in order to build
better PBI and at a more rapid pace but I've never had the time to set it
up. Not what I suggest people to do.... Unless you are a weirdo like me.


One thing I have thought about is trying to create a dummy file system of
sorts. Like taking a dump of a freshly installed FreeBSD system without
ports/packages installed yet. Restore it into another directory and ether
using FreeBSD Jails or chroot create a one-stop drop for making a PBI. 
Pack the whole thing up in an archive (or pbi). Install it, chroot to it,
install the ports tree, compile the necessary software, and compare the
file system hierachies to track down every file that was added by the
port, and rip a copy of the files. Kill the chroot/jail and use those
files to build a PBI of it on the local file system. Then scrub clean the
jail/chrooted file system (rm -rf'ing), so having a virgin system to
install stuff into is as easy as unpacking the archive again.


Figure, maybe build a set of packages for Xorg/KDE (or steal from install
disk) to skip having to build them for every PBI... yet still be able to
get all the files needed.



The downsides of it,

It would take a long time to install every thing, even with a minimal
FreeBSD install archived. Suitable material to use that file system for
the comparison. A collection of suitable Xorg/KDE packages archived to
safe time.

PBI files would be larger then ever...


While the upsides,

A Very complete method of finding every file needed can be accomplished
with few chances of missing files.

Files not needed: most of the freebsd/xorg/kde3 stuff not a runtime/usage
dependency could be stripped easily by a script, e.g. stuff like wall
papers , config files & binaries not needed e.t.c.

PBI could be made strictly self contained, any problems would probably
have been deleting a needed file during the FS Trim. Or the irksome
problems that occasionally chop up with getting libraries working with the
PBI's program.



TerryP,

A Screw ball for doing things the best he can when ever he can.


--
    
Email and shopping with the feelgood factor!
55% of income to good causes. http://www.ippimail.com



More information about the Pbi-dev mailing list