pcplantdb

[Chad Knepp <pyg@galatea.org>]

Hey yall,

Sorry this is so long, but try and read through... good stuff below,
etc.

Lawrence F. London, Jr. writes:
 > <snip>
 >
 > * SUGGESTED NAME CHANGE
 > <*>
 > [BTW I just noticed that ibiblio hosts the "Open Video Project".]
 > 
 > Perhaps the name of our project could be revised this way:
 > 
 > permaculture.info
 > The Open Permaculture Project
 > An open library of plants, permaculture and their relationships
 > <*>

Although I think 'permaculture.info' really says enough, as a
subtitle, I like 'The Open Permaculture Project' the best so far.  I
don't think the 'open library of plants, etc.' could be descriptive
enough when you start to name permaculture elements like 'plants'.  To
continue accurately in this way you would have something like 'An open
library of plants, animals, people, energy, watersheds, construction,
tools, community, permaculture systems, and their
relationships'... kind of a mouthful.  'An open library of
permaculture and its relationships' maybe...

 > > And just to clarify, I offered my server as a working Zope prototype, not
 > > as the official store for all permaculture.info content. Having the Zope
 > > instance hosted anywhere at this point allows for the collection of
 > > Apache/Zope server statistics and use patterns. Collecting this data will
 > > serve as a hard metric for required server capacity planning in future
 > > endevours. I anticipated at least a six month window before ibiblio could
 > > deploy Zope effectively.
 > 
 > I decided to inquire further about this timeframe and emailed Fred about 
 > it, fwiw:
 > 
 > <>
 > -------- Original Message --------
 > Subject: pcplantdb project
 > Date: Sat, 14 Aug 2004 11:24:13 -0700
 > From: Lawrence F. London, Jr. <lfl@intrex.net>
 > Organization: http://market-farming.com  --  Venaura Farm
 > To: help@ibiblio.org
 > 
 > Fred:
 > 
 > The pcplantdb people are trying to decide what db software to
 > use. This will affect where the project is hosted for the long
 > term.
 > 1) Can you say with any certainty, in terms of months or years,
 > when ibiblio will provide a dedicated Zope server for developers to
 > use?
 > 2) Would you, now or later, accept donation of such a server and if
 > so how much would it cost. If yes and there would also be
 > administrative costs, how much would that amount to?
 > 
 > - They already have a temporary Zope server to use for development
 > purposes so if option 1) looks favorable to the project group or 2)
 > is possible and they can raise the money then they can migrate to
 > ibiblio at the appropriate time.
 > 
 > <>

Great!  This was pretty much what I would have asked.  Post their
response to the list when you get it.

 > > I definitely can *not* host all intended content on the scale as infered
 > > in previous posts, especially streaming media.
 > 
 > That would take a LOT of drive space and bandwidth for sure, if things go
 > as desired.
 > 
 > The best option now seems to be that if Zope is the chosen platform
 > then develop the prototype on your machine and wait for ibiblio to
 > make a dedicated Zope server available whereupon we migrate to
 > there from arashi. By all means stay in frequent touch with Fred at
 > ibiblio about this.

It is/was not my intention for John Howe's server to be anything but a
temporary dev platform.  I even looked into crazy ideas like port
forwarding through our DSL router to my laptop for this as well as
asked my ISP friends how possible Zope was in the near future.  Since
I'm currently on a PV power system folks here kind of frown on leaving
a computer on all the time, so I could really only make it accessible
during [my] days.  Skyhouse Consulting is migrating their hosting
business to [none other than] johncompanies within six months or so at
which time Zope would be available on galatea.org.

More on Zope:

Performance: I have said in the past that performance was a possible
issue with Zope and I want to explain why I don't think that is really
an issue for this project.  1) One of the advantages of Zope's OO
design is that it can scale as big as we want it w/Zeo
effortlessly... if adding another server is an effortless sort of
thing.  I also have some crazy ideas about scaling with Zeo on the
client side as well thereby offloading CPU issues to clients.  This
will only work for higher bandwidth clients (cable modem/DSL/etc. not
dialup).  2) Anything can be optimized and we can work around
bottlenecks by profiling and writing critical bits in C or just
creating a better algorithm.  SUMMARY: Zope has the capabilities to
overcome any performance issue.

Why I think Zope is good: I don't think I've really put out why I
think Zope is the best choice for this project so here it is.  1) It
is the all-singing/dancing product but more than that is there are
already a lot of open source products we can incorporate into HG with
little development time invested.  I've been looking at the ZWiki
product a lot lately and we could totally do something like 'class
HGWikiPage extends ZWikiPage', override a few methods where we do
things differently and have an article like object ready to use in
half a day.  I haven't even looked at the more heavy duty
CMF/CMS/plone/silva stuff yet.  I bet there is a moderation system in
some product or another (squishdot?) that we can incorporate in a very
short amount of time.  2) Zope is designed for and provides a lot of
tools (like Interface) to facilitate collaborative development.  It
also has tools (ZMI) that allow for people with less technical
orientation to make significant contributions in development and
maintenance.  3) Geeks, notice I didn't say anything about the OO
throughout bit.  It really isn't that much of a selling point
comparatively.  It was why I started looking at it but not why I'm
sold on it.  SUMMARY: Zope will probably halve the development time it
will take to go as far as I want to go with permaculture.info.  It
will also be easy to maintain and improve post-project by less
technical people.

Lawrence F. London, Jr. writes (8/14 Re: platform):
 > Chad Knepp wrote:
 > 
 > > Ok to add to the tallies we have afaik:
 > > 
 > >   1) Use Zope and 'damn the torpedoes'.  Start on arashi and move to
 > >      ibiblio when they get a Zope server running or buy space at
 > >      johncompanies or some such if ibiblio doesn't.
 > 
 > Find out if and when ibiblio will have a Zope server with or
 > without supplemntal funding from us. If they definitely will (one
 > way or another) within a year then start work on arashi but get the
 > account set up on ibiblio and begin hosting permaculture.info there
 > and not on arashi; simply link to the prototype database at
 > arashi. Continued access to funding and use of the
 > permaculture.info domain would be contingent on this plus the
 > eventual (but within a specified timeframe, like as soon as Zope
 > exists on ibiblio) complete transition of the database to ibiblio.

Looks like you're on this already.

 > <snip>
 >
 > > opinion is that at this early stage of development making the best
 > > choice technically is more important than the short term project
 > 
 > Short term project hosting? I thought we were talking about a (as
 > far as possible) permanent home for permaculture.info, which I feel
 > should be ibiblio.  Technically, of your four options above,
 > without having much technical expertise I still think that 2, 3 or
 > 4 are a better choice than 1) Zope. This is mostly based on input
 > from Rich. An analogy that comes to mind: Zope is to Outlook
 > Express as options 2,3,4 are to Pine or even Mozilla email
 > client. I'm mostly guessing at this but it is my impression anyway.

I think the domain name is something that has far more assurance of
being permanent.  If Bush is re-elected ( ;-) we could spiral into an
economic recession, funding could be cut to ibiblio, and they could
dump us in the street.  Sure this is unlikely but there are really no
guarantees other than what we collectively manifest.  I am still 100
percent behind trying to be a part of ibiblio but regardless we can't
really be certain that it will be the forever home of
permaculture.info.

I have a different analogy for Zope.  I think it is kind of like
XEmacs.  It has a point and click gui that is [sort of] easy to use if
you need it but underneath it is a super powerful highly customizable
platform.  Rich's comments about Zope where largely his first
impression of the point and click gui of Zope called ZMI.  Zope is
really a very serious piece of software with a highly active
development community.  Zope 3 will feature a complete rewrite that
will better performing and easier to develop for.  Zope is to python
what Linux is to Open Source.

Lawrence F. London, Jr. writes (8/14 Re: synthesis):
 > <snip>
 >
 > 2) if answer to 1)a) is a long time from now, i.e. over a year and
 > 2)b) is out of the question Then you (Stephanie), since you favor
 > ibiblio and the software they currently provide, could decide for
 > all of us to do the project there. If noone but Rich likes this
 > option then put out a call for other programmers (PC list and
 > sanet-mg and elsewhere - they're out there and I'll bet you could
 > find all you need within a month) who will work in the environment
 > ibiblio provides and offer them a fair wage. I am firmly convinced
 > that a lot of funding _will_ become available to complete the
 > software permaculture.info uses for its database and maintain it,
 > website and any other software that is used. Adequate funding
 > should become available relatively soon to ensure that good wages
 > can be paid to the programming staff. With hosting on ibiblio that
 > would free up nearly all funding to pay you, PI and the
 > programmers.

I don't mean to imply that I am unwilling to go with something other
than Zope if the group chooses it.  I think the best technical choice
for the project is Zope for the reasons above.  I started Eden with
the same goals I have now, just with less information.  If the group
really wants to go with an RDBMS I still think I'm the best programmer
for your buck... in Missouri anyway ;-).

The consensus process is really about communicating all the issues
involved and although the decision about platform has been fairly
laborious I think we have most of them on the table.  Lawrence feels
very strongly about this issue and has effectively stated that he will
block consensus on this issue if we do not choose ibiblio for our
hosting.  I am reluctant to make a technical decision that is not
based solely on technical merits, but this is not a blocking issue
for me.  I feel like I've expressed my concerns and that people
understand them.

Furthermore, Lawrence, I don't know if I can say this in a way that is
easy to hear, but I'm concerned that you feel this strongly about an
issue and have such a high degree of attachment to a particular
outcome.  Blocking is pretty heavy trip to lay on the group and I hope
that this is a unique situation that you happen to feel particularly
strong about and is atypical for you.  On other decisions we've made
you, for the most part, have been very open to difference and willing
to compromise.  In the future I would appreciate if you could express
your concerns about the decision and allow some discussion before
presenting your decision.  You did do this in your 'Aug 11th Re:
platform revisited'.  This is the kind of info. that is helpful in our
discussions.

Well, I'll let someone else generate a platform proposal.  If 'said
proposal generator' wants to just use my numbers I would like to
recommend 3) over 4).  It doesn't really add much time to the
development and increases portability (could use MySQL or
PostgreSQL...) as well as enables cleaner OO model.

Ack! update: So I've gotten parts of Ack! working in Zope, but for
data entry I much prefer the CLI to a browser interface.  There are
still things I don't understand and I haven't been able to 'reverse
engineer' the object layout in Zope's ZODB instance in Data.fs in
order to have the CLI and the web based interface access the same
objects in Data.fs.  Since we have stepped up the platform discussions
I haven't been working on it as much against the possibility that we
might not choose Zope after all.

Sorry this is taking so long.  We will get going here one of these and
then stuff is REALLY going to happen!  Hang in there y'all.

-- 
Chad Knepp
python -c 'import base64;print base64.decodestring("cHlnQGdhbGF0ZWEub3Jn")'