[IPOL discuss] testing the new version of the demos

[Juan Cardelino]

I'm starting to re-implement my demo (which i've finished in
september) on the new version of the code. There area a few issues I'd
like to point out:
*I couldn't run the demo from ipol_demo_release-20101116.tar.gz at
first. After some hacking, I've discovered that the demo was not
built, because I've just ran ./demo.py. Although it is mentioned
indirectly, I think it should be explicitly said that you first need
to do a ./demo.py build before running ./demo.py, because the default
action is to run (without building) and the error you get is difficult
to read at first.
*Finally, build failed for the cm_fds_mcm_amss demo, because I had a
timeout when downloading the code. This happened from two different
places and at two different times of the day, so I'm sure it is not my
problem.
*I've tried to correct a typo on the "Anatomy of an IPOL demo" but it
is locked. shall i pass you the location of the typo? As you said
"correct it yourself..."
*do we still need to get the .tar.gz? or can we get the code from git?
if so, what is required (ssh access or whatsoever)?
*I still think that we could avoid using hard coded names for the
input images (input_0.png appears 4 times in the saxpb demo), can't we
have some strings named image_0, image_1 and so on. those should be
attributes of the app class. I'm not talking about a mandatory thing,
but maybe if we put it in some demos, people can use the idea.
Thats all for now.
Best,
         Juan

On Mon, Nov 1, 2010 at 10:26 AM, Nicolas Limare
<nicolas.limare at cmla.ens-cachan.fr> wrote:
> Hi everyone,
>
> Lots of things have changed in the demo system during the last 3
> weeks. After too much teasing, everything is now available and
> documented.
>
> # source code
>
> You can find the source code here:
>   http://demotest.ipol.im/dl/
>
> Now the archives are updated every day, with the date information, and
> 3 archives are provided:
> * ipol_demo_lisani : Jose Luis code branch
> * ipol_demo_nil : my code branch
> * ipol_demo_release : the "released" code branch, as seen on
>  http://www.ipol.im/pub/demo/
>
> # documentation
>
> The documentation and tutorials are updated on
> https://edit.ipol.im/admin/demo/. Please tell me about typos and
> missing informations (or, better, correct it!)
>
> # changes
>
> I tried to simplify the code. The version you saw in september was the
> first draft, and I think the design could be greatly improved. Here is
> the list of the current changes:
>
> * layout
>  Now all the demos are in the /app/ subfolder. Everything else is in
>  the /lib/ subfolder. The library code has been reorganised into
>  differents files.
> * build system
>  No more makefile; programs are built with python commands, by the
>  app.build() method. You must write this method when you create a new
>  demo; some help functions are provided in /lib/build.py, to download
>  a source code archive file, decompress it, and run its build script
>  (make, cmake, ...)
> * 3-steps processing
>  After input data and parameters selection, the algorithm is run via
>  3 pages/methods:
>  * app.wait() checks and saves the parameters, shows a "wait" page
>    and triggers a page refresh
>  * app.run() reads the parameters and runs the program, then triggers
>    a redirection
>  * app.result() shows the algorithm result
> * path/url
>  No more self.path() and self.url(), they were the symptom of a bad
>  design. Now all the paths and urls are available as class
>  attributes: seld.key_dir, self.base_url
> * template parameters
>  The app class is always passed as a parameters template (thus you
>  can use app.base_url and app.key). The urld dictionnary parameter is
>  not needed anymore, it had few benefits
> * description
>  Instead of using the app.description attribute, you should now write
>  the algorithm description into an input.html template
>
> # development server
>
> The current situation is not flexible enough. I'd like everyone to be
> able to access the code branches in real time, read detailed
> changelogs, contribute, send and exchange patchs, and test the code
> online.
>
> This would be achieved by a development server, with shell accesses
> and demo instances, but the current demotest server *is* the demo
> server, so I can't risk a server overload. Deploying and installing an
> external development server wouldn't work either, because we need
> some CPU power to run the demos.
>
> We received the next demo server last week (4x Xeon7500, 64
> hyperthread cores). I will install it at the end of November, once the
> lab environment is ready. Then, the current demo server will be free,
> and we will have it available for use as a development server.
>
> --
> Nicolas LIMARE
> http://nicolas.limare.net/                         pgp:0xFA423F4F
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.10 (GNU/Linux)
>
> iEYEARECAAYFAkzOseAACgkQvviFAPpCP09E2wCfaJoADm40yUDqDyUQRotCxREG
> 4BYAoJOCLDD0fvC1hwNz2HKadwIKDLji
> =lcM1
> -----END PGP SIGNATURE-----
>
> _______________________________________________
> discuss mailing list
> discuss at list.ipol.im
> http://tools.ipol.im/mailman/listinfo/discuss
> Spam detection software, running on the system "smtp.fing.edu.uy", has
> identified this incoming email as possible spam.  The original message
> has been attached to this so you can view it (if it isn't spam) or label
> similar future email.  If you have any questions, see
> the administrator of that system for details.
>
> Content preview:  Hi everyone, Lots of things have changed in the demo system
>   during the last 3 weeks. After too much teasing, everything is now available
>   and documented. # source code [...]
>
> Content analysis details:   (5.4 points, 5.0 required)
>
>  pts rule name              description
> ---- ---------------------- --------------------------------------------------
>  1.8 FSL_HELO_NON_FQDN_1    FSL_HELO_NON_FQDN_1
> -1.0 T_RP_MATCHES_RCVD      Envelope sender domain matches handover relay
>                            domain
>  2.0 BAYES_50               BODY: Bayes spam probability is 40 to 60%
>                            [score: 0.4641]
>  2.6 HELO_NO_DOMAIN         Relay reports its domain incorrectly
>
>
>
>