Currently I am considering about delivering a Webware application to a
customer without the source code ot the servlets, i.e. only handing out the
*.pyo files. This may prevent a little bit from fiddling around with the
sources, though I know it is not a real protection since it is possible to
Here is another suggestion for the Appserver script.
It fixes the problem that you cannot start it with startproc, like this
startproc -f -u $WEBKIT_USR -g $WEBKIT_GRP \
-l $LOG -p $PID_FILE $WEBKIT_DIR/AppServer
since there is no way to tell startproc to change to execute AppServer in
the
Here is a suggestion for improvement of the Appserver script:
When debugging an application, I like to run the Appserver with unbuffered
Python output (Python option -u). Also, it might make sense to run it with
Python using optimization (Python option -O). However, you cannot give these
, DirectoryFile has precedence over ExtensionCascadeOrder.
So in the above case, 'index.html' would be prefered over 'main.py'.
Gtx
Christoph Zwerschke
patch of Application.py module:
findDirectoryIndex as in Webware 0.8.1
# Handle directories
if debug: print ' directory = %s' % repr
, too, but I haven't
tried them yet. Try googling for Tsearch2, OpenFTS and PyFTS.
Maybe this could be an alternative to Lupy.
Gtx
Christoph Zwerschke
---
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools
Hi all,
I’m just experimenting with WebKit and database connection pooling, and
some questions/suggestions came up which I hope can be answered here:
As far as I understand I have to use either MiscUtils/DBPool.py which
is pretty simple, or (as I noticed just now by searching the archives)
a
Warren, thanks for the quick reply.
I think we should proceed as follows:
1) The MiscUtils/DBPool.py should be kept as it is, but it should be a
little bit better documented. The advantage of DBPool will be that the
code is very easy to understand. So it may serve as a “starting point”
for
import threading
t = threading.currentThread()
Then you can get and set arbitrary attributes on that
thread to achieve thread-local storage. So a simple
implementation of a function that gets a thread-local
connection could look like this:
...
Thanks, that looks quite reasonable and
to the
mailing list. Of course I will discuss real changes before checking them in.
-- Christoph Zwerschke
---
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles
I went through most of the Webware toplevel doco today and checked in
lots of html cleanup and cosmetic improvements (I hope it's ok to do
this directly in the trunk, since it can hardly break anything).
Actually, the whole documentation creation process should be overhauled.
Currently it's
Actually it was not only DBUtils why I asked to postpone, but I still
hadn't tested the last beta with my main application and had only time
to do it this week. So I thought it would be a better timing to have the
next beta cut one week later.
Also, before the 0.9 release, CVS reminescensces
Ian Bicking wrote:
To me, Webware's lack of a conventional Python installation process
has always been a serious problem. There's no reason to use Webware's
distribution mechanism (which clearly is slow and error prone, which has
a lot to do with the many-year delay of 0.9) when you have a
Problem 1:
The AppServer and AppServer.bat shell skripts do not work if they are
not started from their containing directories. Particularly, I cannot
used startproc to start the AppServer script.
Suggestion for improvement:
Add this to AppServer:
cd $(dirname $0)
Add this to AppServer.bat:
The MakeAppWorkDir script has an option --cvsignore that creates
appropriate .cvsignore files, so that for instance the Cache directory
will be ignored when you store the created directory in a CVS
repository. I think that is a usable feature, but many developers and
Webware itself have moved
The Webware installer copies the stylesheet in the top level Docs folder
to the Docs folder of all subpackages. Does anything speak against
removing that propagation and using relative links pointing to the top
folder instead? This would make the doco and doco creation process a bit
simpler
Chuck Esterbrook wrote:
Agreed, but...
Microsoft Windows.
Hm, what's the problem with Microsoft Windows?
I mean, except Microsoft Windows itself.
-- Christoph
---
SF.Net email is Sponsored by the Better Software Conference EXPO
September
Jason Hildebrand wrote:
I don't think Christoph was suggesting using symlinks in the filesystem, just
to use relative references in the HTML like so:
link rel=stylesheet type=text/css href=../../Doc/styles.css
Yes, I was talking about that. I didn't want to set any links in the
Ian Bicking wrote:
On the general subject of a launch script, I've attached a Launch.py
replacement that I've been using.
Thanks. I will try to integrate this and the other changes until next week.
-- Chris
---
SF.Net email is sponsored
I have just checked in a hack that allows the use of sub-contexts in
WebKit. This gives you more flexibility to arrange and embed your
contexts in an already existing tree of static pages without having
resort to mod_rewrite.
This hack also makes it possible to make the documentation in all
The Admin servlets ServletCache.py and AppControl.py in the Admin
context are broken:
When I start Servlet cache, I get
'Application' object has no attribute '_servletCacheByPath'
When I start Application Control, and click on ClearCache, I get
'Application' object has no attribute
F. Behrens wrote:
I just came across this page:
http://www.turbogears.org
Yes, looks like a modernized version of Webware. Chosing kid as a
component is certainly a good idea. Anyway, I'm planning to work on the
KidKit for Webware component. It's good to have some alternatives to
create
Speaking about the next release, I think it is very important that we
work on the release notes. I will add my own changes soon. If you
remember some things that have been added since 0.8.1, please let me
know or add them to the notes in the repository. For instance, there
must have been some
The servlets ServletCache.py and AppControl.py in the Admin
context are broken...
I have just gone over the admin pages and fixed these things.
-- Christoph
---
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free
Just for the record, since nobody protested, I have re-implemented the
methodNameForAction() transformation for action names. By default it
is the identity, so there will be no compatibility issue.
Because of this, the current behavior is nearly compatible with
OldStyleActions. The only
I noticed that in 0.9beta2 the 404Text.txt page was not supported any
more. It was still there and mentioned in the docs, but not used.
Instead, a standard HTTPException error message was printed.
Was there a deeper reason for this?
Anyway I have re-implemented the old behavior. Only if there
I've just replaced the Unix start script WebKit/webkit by a folder
StartScripts containing start scripts for not only RedHat but also other
flavors of Unix, like SuSE, Debian or NetBSD, plus a generic script that
has been tested under Solaris and should run nearly everywhere.
If you put a
I just checked in a change that makes the PythonServletFactory
responsible for .pyc and .pyo files as well besides .py files.
For the default configuration, this makes no difference,
since .pyc and .pyo are in both ExtensionsToIgnore and FilesToHide
so they will not be served anyway.
But by
I really would like to go for the 0.9 release now.
Mark, if you do not find the time, and nobody else volunteers, I can try
to do it this weekend. Let me know whether I shall do it and which
things to heed besides the ReleaseProcedures.html which seem to be outdated.
I'm suggesting to make
In the new workspace there are 2 problems. The output isn't nearly as
friendly and my command window is completely frozen.
The problem seems to be that the DebugAppServer freezes always when an
exception is raised - in this case, a KeyboardInterrupt exception. In
the old version, the Ctrl-C
The problem with DebugAppServer was that it finished without stopping
the close thread when an exception was raised and not handled. I've
just checked in a fix for the DebugAppServer so that it shuts down
properly. Please check whether it is ok that way.
-- Christoph
Shall we make a 0.9rc first or directly release it as 0.9 and fix any
problems in 0.9.1?
Unfortunately, here in Germany the weekend is already over so I will cut
the release in the course of the next week unless somebody else jumps
in. Today I'll just try to update the ReleaseProcedures.html.
Some things are still unclear concerning the ReleaseProcedures.
* A download mirror at ebad.pair.com is mentioned. I assume it is not
used any more. We have the download directory at w4py.org and we can
also use the SourceForge files section.
* The ReleaseHelper script cuts the release
and in the tag if it is a final release, they should be *copied*
to ReleaseNotes-0.9, and ReleaseNotes-X.Y should be emptied
I meant, in the *trunk*.
---
This SF.Net email is sponsored by the JBoss Inc.
Get Certified Today * Register for a
I'm currently going through the tracker and task items on the
SourceForge page, closing all tasks that have been done, are not
applicable any more or are simly obsolete due to long time.
Do we need all 4 trackers (Bugs, Support Requests, Patches, Feature
Requests) and the task manager? Maybe
By the way, there is also the future.html file in the Docs folder with
open issues. So the to do items are really scattered everywhere.
---
This SF.Net email is sponsored by the JBoss Inc.
Get Certified Today * Register for a JBoss Training
In order to consolidate things a bit, I have now
closed/disabled the following on SourceForge:
- the Support Request tracker:
has never been used anyway, use webware-discuss instead
- the task manager:
has not been used since years;
I have copied ideas and tasks which were still open
to
:
http://www.webwareforpython.org
Please report possible problems to [EMAIL PROTECTED]
Include all appropriate version and platform information including
Webware, Python, web server, op sys, browser, database, etc.
The scheduled release date for the final relase is November 12, 2005.
-- Christoph
I am running a Debian variant and ran update-rc.d to create the proper
rcX.d entries for my run levels. These are symbolic links to the
symbolic link in /etc/init.d. The problem is that now the startup script
reports /etc/init.d as my $WORK_DIR
Good you found that problem, Seth.
You're
By the way, I also thought about adding a generic Webware Posix start
script that is not a shell script, but a pure Python script. Would that
be a silly idea? I think it should work, using the shebang mechanism.
-- Christoph
---
This SF.Net
Actually, install.py should have copied the Debian script automatically
to WebKit/webkit if /sbin/start-stop-daemon exists (can you confirm that
this also exists in Ubuntu?), but I just saw there was a typo in the
check for start-stop-daemon. I have fixed that too.
-- Christoph
I checked out the latest svn and for some reason the Generic start
script still isn't working for me (I swear it was working for a bit ;)
I think I found the problem. Python's readlink() can return a relative
path, so a simple recursion will not work when the links cross directory
boundaries
Don't worry, that checking looks only so large because one file (in
Experimental) was converted from spaces to tabs (Webware/your standard)
and every line is reported as changed. The rest is mostly white space,
word wrapping (I think it makes sense to word wrap at max 80 chars). I
didn't
BTW, spaces around comparison operators make pylint happy.
Anyway, you're completely right, this was not the moment for such
clean-up and nit-pick because I unnecessarily forced others to review
many petty and unnecessary things. I'm sorry.
-- Chris
I understand your frustration. 0.9 has really been delayed too much.
I'm still aiming at Nov 12 for the final release. I won't touch the
source except for real issues. So far only two minor problems have been
reported which have been fixed already. So I don't think we need a rc2.
And the
Is it ok if I cut the final 0.9 release this weekend? I heard no
complaints about the 0.9 rc1 in the last time.
-- Chris
---
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. Download
it for
I cannot break out of ThreadedAppServer with Ctrl+C whether at the
command prompt or running inside WingIDE.
The DebugAppServer can break out fine.
I'm on Win XP Pro with Python 2.4.1.
It works on my Win XP Pro and Py 2.4.1 (tried only command line
(cmd.exe), not WingIDE), and in the code,
Just checked the current trunk under WinXP + Py2.4.1 and I'm still
experiencing no problems breaking out from th app server with Ctrl/C.
So I will proceed and release 0.9 today. According to Murphy we will not
find the real bugs until shortly after the release anyway ;-) But since
the release
Or should I update the w4py.org server to 0.9 before making the release
in order to have another real-world test?
-- Chris
---
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. Download
it for
Regarding w4py.org, I'm not sure what specific security issues you're
referring to with regards to having the whole tree available. As long
as the .py files and other scripts are not executed by the web server,
we'll be fine.
They ARE executed, that's what I'm talking about. For example,
Ok, I've fixed all that and updated the w4py.org server already. It now
runs on Webware 0.9, and the docs are served directly by the webserver.
The new releases and online docs are also available via SourceForge.
-- Chris
---
SF.Net email
Webware 0.9 has been released.
The new release includes numerous enhancements, additions and bug fixes
over the previous release. We can list only a few of them here:
* easier installation
* improved documentation
* improved examples
* bug fixes
* a built-in HTTP server for immediate playing,
The ReleaseNotes for 0.9 say AutoReload no longer catches servlet
updates (since servlets can be reloaded without restarting the entire
server anyway). (This was checked in by Ian in Revision 1956).
Generally, I'd say this is a good idea, but I noticed there is also a
drawback: The exception
My impression was that it was a bug in Python since the servlet's
module *is* getting reloaded. Maybe if you can recreate the issue
in a small standalone example, you can ask comp.lang.python for help.
Thanks for pointing me into the right direction.
I checked this and you're right. The
If it's fixed in Python 2.4, is simply upgrading an option?
I have checked that again, and yes, it has been fixed in the traceback
module a year ago and thus should be available since Python 2.4:
http://cvs.sourceforge.net/viewcvs.py/python/python/dist/src/Lib/traceback.py?r1=1.30r2=1.31
BTW, is it by intend that the Webware mailing lists do not have a
reply-to header? I always forget to fix the recepient, since a reply to
a posting in the list goes to the author by default, not to the list.
Can this be changed?
-- Christoph
Seth Remington wrote:
Reply to All is a quick userland fix.
Yes, but I always forget it because in other lists this is not needed.
Also, this results in sending the message to the list plus to the author
who will receive it two times.
-- Chris
Frank Barknecht schrieb:
You can normally disable this behaviour (getting list mails twice)
How? I only know norcv and hide which will do something different.
And even if I could disable it for me, I cannot be sure others have
disabled it.
so at least for me it is usual to press L in Mutt,
Frank Barknecht wrote:
I already feared it is an old and controversial issue again... ;-)
If you google for the keywords I mentioned you can read about the
issue in all its glory. Of course, the Thunderbird bug discussion also
went off-topic to fight that old fight. ;)
You seem to be right.
Chuck Esterbrook wrote:
It's been so long since I updated the web site, I don't recall how
we're doing it. Can someone who does make these tweaks, or post on how
we're set up?
It's in SVN/Webware/HomePage/Home/Papers/Templates/Packages.csv.
You can change the CSV file in the repository, and
You can change the CSV file in the repository, and then make an update
on the webserver.
It was not completely automatic. The table on the page was hardcoded; I
needed to change that to include the data from a file which has to be
created by running gen.py once after changing the CSV file.
I just noticed that the name www.webwareforpython.org cannot be found in
the DNS any more. The alias www.w4py.org still works, but since most
links go to the full name, and since there is a redirection to the full
name, this does not help us.
You can check it here, for instance:
Jason Hildebrand wrote:
I have just renewed webwareforpython.org for another two years. DNS
should be back soon.
Thanks a lot, Jason. That was really quick! It's already available again
here from Germany.
-- Christoph
---
This SF.Net
I have just implemented a solution to the ImportSpy problem.
ImportSpy used ihooks to track imported modules, in order to reload
those modules with AutoReloadingAppServer. The problem was that using
ihooks could raise problems if you used Cheetah templates or imported
modules like Kid from
As promised, I have cut a new beta release. It contains all bug fixes
and enhancements made after version 0.9 so far, plus a couple of other
improvements and bugfixes I made this weekend.
Particularly, ImportSpy now uses new import hooks instead of ihooks (if
available) and can be disabled
Mark Phillips wrote:
FWIW, today I downloaded the 0.9.1b1 and installed it on Mac OS X 10.4.6
running python 2.4. Ran through the tests without a hitch. Set up a
working directory, added my app files, and fired it up also without a blip.
That's an important feedback since I've run and tested
I can allow running on an existing working dir, but how should
MakeAppWorkDir behave if the config files or the launcher script
already exist? Overwrite or keep (there could be some customization
in the config files and launcher script already)? Maybe we can add
an option --overwrite, but
In preparation of a new Webware bugfix release, this weekend I have been
processing the Sourceforge bug tracker and implemented some very old
bugfixes and patches, and closed outdated issues. If you have other
changes that shall go to the next release, let me know.
-- Christoph
Can I remove the SelectRelease module in the WebKit package? It seems it
was only used by the AsyncThreadedAppServer which has gone already.
-- Christoph
-
Using Tomcat but need to do more? Need to support web services,
I noticed that after July 5, there are no checkin notificatons on
[EMAIL PROTECTED] any more, though I have
made some checkins to the Webware repository recently.
I have already looked on the server but found no changes in the config
or error messages. Any ideas?
-- Christoph
Chuck Esterbrook wrote:
On 8/11/06, Christoph Zwerschke [EMAIL PROTECTED] wrote:
Can I remove the SelectRelease module in the WebKit package? It seems it
was only used by the AsyncThreadedAppServer which has gone already.
No objections here.
There are also WebKit/Experimental and WebKit
Ian Bicking wrote:
That's odd, I'm getting email from some locations from that same box.
Are you making checkins someplace that is configured to send to that
address?
Yes, but I think I have found the problem: The exim process did not run
any more. I found the following in the messages
I think I'm now done with this round of bug fixing. Webware should now
install and run with all Python versions from 2.0 to 2.5 as claimed by
the Docs and the Property files (at least all the unit tests pass); and
I tested it on some more platforms.
The 0.9.2 beta release will be available in
As announced, a new beta release of Webware for Python is available for
testing. It contains a lot of bug fixes and improvements, including many
old issues from the SourceForge tracker. Backward compatibility with
Python 2.3 has been fixed in many places. The idea is to switch to
Python 2.3 or
Thanks for wanting to share your contributions, Dan.
Got the trunk checked out of svn, but when I then ran install.py, it
modified all the docs and configs, which is confusing the heck out of
svn (svn status gives a host of !'s and M's).
The installer creates html files from phtml files by
Dan Milstein wrote:
I've added java servlet-style filters to Webware, and would like to
contribute the code to the project, if people are interested.
...
If people are interested, I can write up some documentation.
That's surely interesting, thanks.
I'll have a look at it and I'd also like
Webware 0.9.2 has been released.
This release of Webware for Python includes a couple of fixes
and improvements of WebKit, MiddleKit, MiscUtils and WebUtils
(see the respective release notes).
Webware for Python is a suite of Python packages and tools for
developing object-oriented, web-based
Fionn Behrens wrote:
Looks like webware tries to load a servlet file from a dedicated
directory for each psp page. This is a requirement that wasnt there
before and I wonder what it is good for! Would someone please try to
explain? Since a lot of requests on my server go to /virtual/
Fionn Behrens wrote:
From which version did you upgrade to 0.9.2?
So, in consequence, all sources say Version X.Y - I'd guess it should
be 0.8.9 or 0.9.0
That looks like one of the early beta versions between 0.8.9 and 0.9.0
where the version number was not properly set.
No, I did not
Fionn Behrens wrote:
I have a context Servlets (outside webroot) defined for the Servlets I
use from my psp pages. The psp pages are located in the normal webroot
and load via mod_webkit and PSP-Handler.
Ok, I could now reproduce the problem. It happens when you are using
psp-handler with psp
Fionn, can you give me a quick confirmation whether that problem has
been solved with the small patch that I have suggested?
-- Christoph
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff
Dan Milstein wrote:
After looking carefully at HTTPHandler, I think I've found a way
to avoid holding that file in memory. I wanted to ask the list if
there was a reason for the current design that I'm missing. If not,
I'd propose the patch below as an improvement to the built-in HTTP
Just took a look at the fix in svn -- one thing: on line. 36, you're
deleting 'Content-Type' from the headers, and I think you want to be
deleting 'Content-Length'.
Good that you looked over the patch. I've already fixed the lapse.
-- Christoph
Hi Ezio,
your ideas about implementing failover connections and load balancing
sound interesting. Can you email me your suggested patch? Concerning
load balancing, we probably need to somehow discern master and slave
databases and offer a way to request master and slave (usually read
only)
Hi Nick,
you recently posted the following on the Webware mailing list:
I have two webware apps running on the same domain, one on /default/ and
the other on /admin/. To avoid the session cookies from colliding, I
made a small change in HTTPResponse.py:recordSession. The line
Thanks for the feedback, Nick.
It turned out that my original idea didn't quite work in practice,
and then I found that webware already includes the feature I needed.
In http://svn.w4py.org/Webware/trunk/WebKit/Application.py
Application.__init__ uses a setting called SessionName
to control
Thanks, Andrew. I will examine this and fix it before the 0.9.3 release.
-- Christoph
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
On webware-discuss I already announced that I want to release the next
Webware version 0.9.3 very soon. If anybody has suggestions or patches
or bug reports that I should take into account, please let me know.
What about the idea of releasing a 1.0 version without much change if
the 0.9.3
Chuck Esterbrook schrieb:
What about [EMAIL PROTECTED]'s suggestion:
What if you stayed with tabs and then added a converter, so that you
can quickly convert the tabs to the number of spaces that you want?
In fact, this might be a good idea for a lot of free source Python
projects
I hadn't expected the issue is still so controversial ;-)
It still would be inconvenient, because you always have to think about
this issue and either convert back and forth manually, or use a script,
or change the settings in your editor. I'm working with a lot of Python
code from various
I recently stumbled over the following lines in ThreadedAppServer:
def makeInput(self):
return self._sock.makefile(rb,8012)
Is there any reason why the number 8012 was chosen? I assume the intend
was to use an 8192 Byte (8k) buffer, and this is just a typo, or am I
Chuck Esterbrook wrote:
On 3/23/07, Christoph Zwerschke [EMAIL PROTECTED] wrote:
I recently stumbled over the following lines in ThreadedAppServer:
def makeInput(self):
return self._sock.makefile(rb,8012)
Is there any reason why the number 8012 was chosen? I assume
Chuck Esterbrook wrote:
All sounds good except I wonder if making it relative to threads is
worthwhile the level of indirection. We can just default it to twice
the size. But I leave it up to you. Maybe there's something I'm
missing.
I wanted to explicitly set it to something in the default
Chuck Esterbrook wrote:
I think they are already documented in the regular docs. I'd prefer to
just give a pointer to that if we don't have one already. As someone
who already knows the settings, I could be biased in this matter.
There are currently two places where things are officially
One thing I always disliked about Webware is the static 404 error page.
First, the name '404Text.txt' is hard coded and misleading because it is
a complete HTML page (maybe it has only been the body part in older
versions, as it is still wrongly documented), so it should better be
called
Jean-François Piéronne wrote:
Very good idea, allowing declaration of a default error page will be
very useful.
May be it would be also interesting to allow the declaration of
HPPTErrorPage for application raised exceptions. For example:
HTTPErrorPage = {
'NotFound':
Just a short note for those who waited for the 0.9.3 release.
Unfortunately, it will take some more time since I didn't find enough
time to complete it this weekend and there are still some important
issues on my list. Plus, I just noticed a new issue with Python 2.5 and
AutoReload (the
Hi Ben,
I'm going to implement your patch, but I'd like to cut down the number
of additional config settings. The number of parameters is already
scaring enough ;-) What do you think about:
1) Assuming that the class name is always the same as the module name
(this is also backward compatible
Ok, this is now in the trunk if you want to check it out.
Ben Parker wrote:
Agreed about too many config settings - I thought about having the
Session / SessionStore set through some kind of runtime configuration.
Allow each context to set its own session/store combo perhaps? but that
is a
Webware 0.9.3 has been released.
This release of Webware for Python includes a couple of fixes and
improvements of WebKit and some cleanup of the overall Webware codebase.
Please have a look at the WebKit release notes for details.
Webware for Python is a suite of Python packages and tools for
Ezio Vernacotola wrote:
I need a database connection pool that can reconnect to a different
replicated failover server if the main server stop working. I found
DButil SteadyDBConnection works great in transparently reconnect if the
connection is lost, but to try a different server I'm
1 - 100 of 140 matches
Mail list logo
