Webware for Python

Hi,

I have a rather big ASP/IIS-based system which I'm trying to rewrite and 
  extend with python (more specifically, using Cheetah/WebKit).

I've already wrote some useful code in Cheetah/WebKit and now I'm 
thinking about the best way to integrate it back into the legacy system.


Basically I see the simple route:

Put Response.Redirect('webkit/..') into .asp pages. But I need to shared 
data between IIS and WebKit I'm not sure how to do it. First of all, I 
'd like that both env. match the session state: create WebKit's session 
when IIS session starts and end WebKit's session when IIS session ends. 
Would appending parameter _SID_=xxx start the session? If yes, then how 
could I end it? Add another redirect to the page that would call 
self.session().abandon()?

Other ideas for integration and discussion are wellcome.

JZ [mailto:webmaster@...] wrote:
>
> Aaron Held <aaron@...> napisal(a):
>=20
> > Is odbc dbapi compliant?
>=20
> I suppose, it is. But I do not know how much.

The accepted wisdom is that with the Windows odbc module, your mileage =
may
vary. mxODBC is what you really want, if you need a DB-API compliant =
ODBC
module.

> > I am pretty sure that DBPool needs a dbAPI module to work with.
>=20
> So what to do with it?

I suppose you need to do a bit of debugging. Have a look in DBPool to =
see
what it does with the keyword arguments you specified, for example.

Paul

I have to change connection of my Webware application from MySQL to
MSSQL2K. I am using DBPool module.
For MySQL I use:

import MySQLdb
datapool[dbName] = DBPool(MySQLdb, 5, host='localhost', user='user',
  passwd='pass', db=dbName, compress=0, cursorclass=MySQLdb.cursors.DictCursor)

For odbc I tried to use:

import odbc
datapool[dbName] = DBPool(odbc, 5, dsn='myDSN')

but it does not work. :(

And another question. Is there a possibility for odbc to accessing
database results by fields? I cannot find good documentation for odbc
module. The only web page I found is very short. :(

What is the correct syntax for MiscUtils.DBPool and odbc?

-- 
JZ

Hallo,

I'd like to add a CSS-class statement to a form generated by
FunFormKit. I thought, that the Field argument "extraHTML" would do
that, but somehow it won't. I'm testing this with the Register.py
example, whose FormDefinition I changeed like: 

formDef = FormDefinition(
	"Register",
	[TextField("firstName", maxLength=20,  extraHTML='class="babyblue"', validators=[Validator.NotEmpty()]),
 ...


But my babyblue-class never shows up in the rendered form. Is this a
bug or should I specify the class in another way?

ciao
-- 
 Frank Barknecht                               _ ______footils.org__

Hello,

  I'm using Webware-0.8.1. I would like to make some patches to Webware
(at least UNIX sockets). Which release should I use to do patches? I was
told by friend, that the best one is CVS version, but I don't know
whether it is stable enough (and compatible with stable releases).

  Which one should I do the patches for?

  Some ideas to implement:
- UNIX sockets (I can do this)
- ErrorDocument handler like in Apache. I think that 
  each context should have own error handler (this seems to be harder,
  but it can be done).

  Have a nice day

Věroš Kaplan
--
Věroš Kaplan <veros @ tac . cz>
system disaster 
Tacoma Computers, Staňkova 18a, Brno, CZ
--
||
"A bus station is a place where a bus stops.
 A train station is a place where a train stops.
 There is workstation on my table."

Rather than just sending email, you should post this as a patch on
SourceForge.  Someone will eventually review it and if it looks good, will
check it into CVS.

Sending patches by email isn't a good idea -- it's too easy for them to get
lost and there's no way to assign a particular developer to look into it.

Also, it would be helpful to upload a unified diff (diff -u) or context diff
(diff -c) of the whole file instead of just a code snippet.

- Geoff

webware-discuss@... wrote:
> Can someone reply to this. I depend on it and it's getting old now.
> 
> --> Friday, August 29, 2003, 12:09:38 PM,
> webware-discuss@... wrote:
> 
>> req.serverSidePath accepts relative site paths fine, but does not
>> work with an absolute site path.
> 
>> [snips]
> 
>> res.write(req.serverSidePath('/index.psp')+'<br>')
>> # it returns '\index.psp' instead of the expected absolute path
> 
> This is my fix... (with tabs to match HTTPRequest.py)
> 
> #
> ---------------------------------------------------------------------
> 
>         # kaishaku fix
>         def serverSidePath(self, path=None):
>                 """     Returns the absolute server-side path
> of the request. If the optional path is passed in, then it is
> joined with the server side directory to form a path relative
> to the object.
>                 """
>                 if not hasattr(self, '_serverSidePath'):
>                         app = self._transaction.application()
>                         self._serverSidePath,
> self._serverSideContextPath, self._contextName =
> app.serverSideInfoForRequest(self)
>                 if path:
>                         if path[0]=='/' or path[0]=='\\':
>                                 return
> os.path.normpath(self._environ['DOCUMENT_ROOT']+path)                
>                                 else: return
> os.path.normpath(os.path.join(os.path.dirname(self._serverSide Path),
>                 path)) else:
>                         return self._serverSidePath
> 
> #
> ---------------------------------------------------------------------
> 
> -Kai
> 
> 
> 
> -------------------------------------------------------
> This SF.net email is sponsored by OSDN developer relations
> Here's your chance to show off your extensive product knowledge
> We want to know what you know. Tell us and you have a chance
> to win $100
> http://www.zoomerang.com/survey.zgi?HRPT1X3RYQNC5V4MLNSV3E54
> _______________________________________________
> Webware-discuss mailing list
> Webware-discuss@...
> https://lists.sourceforge.net/lists/listinfo/webware-discuss

I'm planning on using the ExtraUrlPath information to provide a lot of =
services out of a single servlet.

So instead of a swarm of small servlet instances I'll have a small pool =
of very fat servlets.

I'm wondering what implications this has for performance, scalability =
etc.

In order to handle a large concurrent load I'll probably need to enlarge =
my servlet pool - that has some memory implications but there should =
only need to be one copy of the servlet code in memory so the memory =
requirements are all about the data the servlet loads up and shunts =
about.

Importing a lot of packages won't be a problem if done under a =
lazy-loading scheme that only loads them when needed - so the servlet =
startup time shouldn't be a big issue.

So what else am I missing? If there's a big roadblock ahead I'd rather =
know sooner than later.

Ian Sparks wrote:
> I'm planning on using the ExtraUrlPath information to provide
> a lot of services out of a single servlet.
> 
> So instead of a swarm of small servlet instances I'll have a
> small pool of very fat servlets.
> 
> I'm wondering what implications this has for performance, scalability
> etc. 
> 
> In order to handle a large concurrent load I'll probably need
> to enlarge my servlet pool - that has some memory
> implications but there should only need to be one copy of the
> servlet code in memory so the memory requirements are all
> about the data the servlet loads up and shunts about.

You might not necessarily want to expand your thread pool too much.  When
your application is being hit with maximum load, if you double the number of
threads, you may also just be doubling the average amount of time it takes
to process each request, and therefore your throughput won't be improved.
In fact, due to the extra memory usage, you may well see _worse_ performance
if you use a larger thread pool.

There are really only 2 reasons to use more than one thread:

- To allow the app server to do more work on a different thread while
blocked waiting for I/O, SQL query results, etc.
- To allow quick requests to be serviced while a longer-running request
continues to be processed.

I think these benefits can be realized with a relatively small thread pool.

> Importing a lot of packages won't be a problem if done under
> a lazy-loading scheme that only loads them when needed - so
> the servlet startup time shouldn't be a big issue.

Each servlet module only gets imported once, so regardless of the size of
your thread pool, you shouldn't have to worry about the time taken by
imports except when starting the appserver.

- Geoff

...check if webware is sending the right MIME type to
the browser.

this one hit me recently: i was trying to send XUL
file to mozilla, and the browser was refusing to corretly
parse the file. "i added the mime type to apache!" -
i told myself, and infact requesting the file with
apache (bypassing webkit) worked as expected.

so i dig into webkit sources and found that webware
use mimetools "guess_type" function to get the correct
mime type of the file.

python's mimetools has a  "Lib/mimetypes.py" file where are
stored the recognized  types. obviously the XUL mime:

application/vnd.mozilla.xul+xml

was not in the list. so i added the mime to mimetypes.py
file and all started to work correctly.

it would be cool if additional app specific mime types
ìcan  be specified inside Application.config file. i think it's
not always feasible to add mime to python enviroment (think
about web hosting scenarios).

hope this helps.

later,
deelan

Can someone reply to this. I depend on it and it's getting old now.

--> Friday, August 29, 2003, 12:09:38 PM, webware-discuss@... wrote:

> req.serverSidePath accepts relative site paths fine, but does not work
> with an absolute site path.

> [snips]

> res.write(req.serverSidePath('/index.psp')+'<br>')
> # it returns '\index.psp' instead of the expected absolute path

This is my fix... (with tabs to match HTTPRequest.py)

# ---------------------------------------------------------------------

        # kaishaku fix
        def serverSidePath(self, path=None):
                """     Returns the absolute server-side path of the request. If the optional path is passed in, then it is joined with the server side directory to form a path relative to the object.
                """
                if not hasattr(self, '_serverSidePath'):
                        app = self._transaction.application()
                        self._serverSidePath, self._serverSideContextPath, self._contextName = app.serverSideInfoForRequest(self)
                if path:
                        if path[0]=='/' or path[0]=='\\':
                                return os.path.normpath(self._environ['DOCUMENT_ROOT']+path)
                        else:
                                return os.path.normpath(os.path.join(os.path.dirname(self._serverSidePath), path))
                else:
                        return self._serverSidePath

# ---------------------------------------------------------------------

-Kai

Hi all,
sorry for this really dumb question, 
but I'm having this strange situation.

1. I've a vanilla servlet with a funformkit-driven form:
if I'm using mozilla 1.2.1 or 1.4.1 or konqueror (KDE3.1/RH9.0)
to play with it (i.e. submitting something), it's all o.k.

2. I've upgraded to mozilla 1.5 (and firebird 0.7),
and _now_ I see double requests on the appserver log
(like if I ask for the page twice)

3. If I copy the Welcome.py sample page to my working dir
all is fine. Even if I copy the standard AddressList.py
funformkit demo page on my working dir all it's ok.
So actually I've not a standard sample to reproduce the
problem: it happens only with my code (and with the latest mozilla,
of course).

4. I can see the double request looking at the _SID_ cookie,
that is'd placed twice (with Moz 1.5), once with the other
browsers.


Of course I suspect it's something wrong with my configuration,
but since with Moz < 1.5 I cannot reproduce the problem, I'd like
to know if there's someone out there kind enough to give me
an hint.


Best regards to all,
thanks for every advice.
/gp


-- 
gian paolo ciceri - japanman
suddenthinks, excellence through refactoring
++39 340 7218101
gp.ciceri@...
gp.ciceri@...

Victor Ng wrote:
> Have you folks seen Webware crash when a syntactically 'bad'
> servlet is
> loaded?
> 
> If I have a subclass of HTTPServlet and I try to use an undeclared
> variable in __init__, the whole Webware server comes crashing down.

I can't say that I've ever seen it, but it wouldn't surprise me.  Most of my
__init__.py's are empty.

> I've got a simple patch, but I was wondering when/if the next Webware
> release is scheduled?  I'm starting to do a lot of work using Webware
> and maybe I could start contributing patches to the CVS tree?

There isn't a schedule for a next release, but if we get a volunteer to put
in the work to do another release, then it will happen.  There hasn't been
very much Webware development recently, and we could use some new
developers.  If you want CVS access, then take a look at the guidelines at
http://webware.colorstudy.com/twiki/bin/view/Webware/DeveloperGuidelines .
Otherwise, the standard way to contribute is to submit patches to
SourceForge then hope that one of the developers eventually looks at it.

In particular, we could really use someone to just review the existing
patches and bug reports on SourceForge.

By the way, the webware-devel list is probably a better place for
discussions like this.

- Geoff

Hi all,

Is there a better way to force a trailing '/' in the URL of a 
particular servlet than this code:

# ---------- MyPage.py

from WebKit.Page import Page

class MyPage(Page):

	def awake(self, transaction):
		'''Force a trailing '/' at the end of the URL'''
		Page.awake(self, transaction)
		if self.request().uri().split('/')[-1] == self.__class__.__name__:
			self.sendRedirectAndEnd(self.__class__.__name__ + '/')

--T