When attempt to browse my Mercurial repositories it usually takes several refreshes before the repository list is displayed. The configuration is as follows:

  • Windows Server 2003 (Dedicated machine hosted by http://www.server4you.com/.
  • Site has anonymous password protection with self-signed SSL.
  • Mercurial 1.5.3
  • Python 2.6.5
  • Python for Windows 32 extensions 214 py2.6
  • isapi-wsgi 0.4.2

The repositories are being served via ISAPI using the standard hgwebdir_wspi.py file (copy to follow).

Also before doing a clone/push/etc I have to browse the repositories first otherwise hg on my local machine can not locate the site.

What can I do to start tracking down this problem?


# Configuration file location
hgweb_config = r'C:\Public\Mercurial\WebSite\hgweb.config'

# Global settings for IIS path translation
path_strip = 0   # Strip this many path elements off (when using url rewrite)
path_prefix = 0  # This many path elements are prefixes (depends on the
                 # virtual path of the IIS application).

import sys

# Adjust python path if this is not a system-wide install
#sys.path.insert(0, r'c:\path\to\python\lib')

# Enable tracing. Run 'python -m win32traceutil' to debug
if hasattr(sys, 'isapidllhandle'):
    import win32traceutil

# To serve pages in local charset instead of UTF-8, remove the two lines below
import os
os.environ['HGENCODING'] = 'UTF-8'

import isapi_wsgi
from mercurial import demandimport; demandimport.enable()
from mercurial.hgweb.hgwebdir_mod import hgwebdir

# Example tweak: Replace isapi_wsgi's handler to provide better error message
# Other stuff could also be done here, like logging errors etc.
class WsgiHandler(isapi_wsgi.IsapiWsgiHandler):
    error_status = '500 Internal Server Error' # less silly error message

isapi_wsgi.IsapiWsgiHandler = WsgiHandler

# Only create the hgwebdir instance once
application = hgwebdir(hgweb_config)

def handler(environ, start_response):

    # Translate IIS's weird URLs
    url = environ['SCRIPT_NAME'] + environ['PATH_INFO']
    paths = url[1:].split('/')[path_strip:]
    script_name = '/' + '/'.join(paths[:path_prefix])
    path_info = '/'.join(paths[path_prefix:])
    if path_info:
        path_info = '/' + path_info
    environ['SCRIPT_NAME'] = script_name
    environ['PATH_INFO'] = path_info

    return application(environ, start_response)

def __ExtensionFactory__():
    return isapi_wsgi.ISAPISimpleHandler(handler)

if __name__=='__main__':
    from isapi.install import *
    params = ISAPIParameters()


/ = C:\Public\Mercurial\Repositories\*

allow_archive = bz2 gz zip      ; Allows archive downloads.
allow_push = ########       ; Users that are allowed to push.
  • The behavior you describe is strange... I've never seen Mercurial behave like that. I've asked people on G+ if they can help. If I were you, then I would write mercurial@selenic.com to see if there's someone there that can help debug this. – Martin Geisler Dec 22 '11 at 20:10

It sounds like IIS 6 is caching your web pages (you didn't define if you were using Apache or not, so I assumed as it's a Windows server)

Use this link from Microsoft and set the site to Expire Immediately.

|improve this answer|||||

Something caches on the way. Use curl or wget to get the page and check the http headers. Is it better without ssl?

|improve this answer|||||

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.