On Feb 6, 2009, at 11:52 PM, fred2 wrote:
i'm revisiting fipy after a several month hiatus.
previously i've used fipy trunk (~2.0a1) for some needed features.
it looks like there's been an api chg, e.g., my prev. test cases
vnlr.3.tar.gz (emailed ~28may08) still run correctly under STABLE,
but (run &) give some incorrect results under current trunk.
STABLE and CURRENT point at version 1.2.3. Rather I should say that
they *used* to point at version 1.2. We deleted both tags yesterday
because we didn't maintain them and they therefore weren't useful. We
now believe that people should either track a specific branch or tag,
or should run trunk. We're trying to do all significant development on
branches parallel to trunk, only merging when they pass all the tests,
so trunk should rarely be broken (and we now run a Bitten build slave
so that you can see whether the revision you're checking out passes
its tests before you check it out: see http://www.matforge.org/fipy/build)
.
The version string "2.0a1" is unfortunately not very useful, because
we've been lazy and haven't changed that number in about a year-and-a-
half. It is on my to-do list to get svn to incorporate the revision
number automatically.
Anything with version 2.x does, indeed, have a different API from
version 1.x. I'll look over the code you sent last week to see what
changes are needed to bring it up to date.
i'm trying to track down the chg (& refresh my old brain) via the
docs, and assume the trunk is still more bleeding edge than release.
but the website has some inconsistencies. also, i note also the
website mentions that python 2.5 hasn't been tested yet, so maybe
there are some other potential issues for me.
We actually hope to release trunk in a matter of days. Although it's
the point of active development, it's what we use for our own
research, and you can see from the Bitten display that it's quite
robust. You're absolutely correct about the inconsistencies on the
webpage. We expect to get that all sorted out *very* soon.
in any case, if there is an alternate source for the latest docs
(hopefully pdfs of manuals, as
latex of trunk/documentation/ complains on my end, but incipient
trunk release notes also of interest), could you let me know. (no
rush, i'm offline for 2 weeks).
We should have made the new release, with revised manuals and
webpages, long before then.
(1) on home page
http://www.ctcms.nist.gov/fipy/
(Last updated: Wednesday, 05 November 2008 10:43:45 EST)
Even if you don't read manuals...
...please read the Installation Guide and Chapter 6: "Frequently
Asked Questions" of the Manual
(1a) "Manual" link on (1) above (http://www.ctcms.nist.gov/fipy/download/fipy.pdf
) is dead.
result is:
Trac Error
No handler matched request to /fipy.pdf .
Thanks for the catch. I've fixed the link (and a few others like it).
(1b) what's new is too new? v. 2.0a1 not 'STABLE' yet? (see (1c)).
What's new in version 2.0a1
The significant changes since version 1.1 are:
The "version 2.0a1" bit was a mistake on my part. Fixed now.
(1c) "Installation Guide" link in (1) above (http://www.ctcms.nist.gov/fipy/installation.html
) has incorrect version / tag details for 'obtaining fipy'?:
:
This will download a fairly stable version of FiPy (somewhere
between version 2.0a1 and HEAD). If you prefer, you can download
FiPy version 2.0a1 (recommended) with:
Got it. Same problem as above. I mistakenly updated the web pages from
trunk at one point. Fixed now.
