pyff should stop processing if one input MD is missing

[salaun-urennes1]

We are using pyFF to agrgegate our federation metadata with eduGAIN metadata. We had an issue that made us publish the aggregate metadata with eduGAIN metadata only.

After some investigation I found out that pyff does not raise an exception if an input MD cannot be loaded. I suggest this not only adds an error log but raise an exception to prevent an uncomplete output file from being generated.

Here is a sample pipeline where one input metadata file is missing:

• load:
  • /var/www/html/edugain/sps-edugain-metadata.xml
  • /tmp/unknown-metadata.xml
  • select
  • finalize:
    Name: https://federation.renater.fr/
    cacheDuration: PT1H
    validUntil: P6D
  • publish: /tmp/test-metadata.xml
  • stats

Here is pyff output:
None
None
ERROR:root:Don't know how to load '/tmp/unknown-metadata.xml' as None verify None via None
[('file:///var/www/html/edugain/sps-edugain-metadata.xml', None, None, None)]
[('file:///var/www/html/edugain/sps-edugain-metadata.xml', None, None, 0, None)]

ERROR:root:[Errno 1] Operation not permitted

total size: 2
selected: 123
idps: 0
sps: 123

It seems that this part of builtins.py should be changed:

    else:
        log.error("Don't know how to load '%s' as %s verify %s via %s" %
                  (url, params['as'], params['verify'], params['via']))

[leifj]

This is verified in HEAD?

[salaun-urennes1]

Le 08/07/14 15:24, Leif Johansson a écrit :

This is verified in HEAD?

—
Reply to this email directly or view it on GitHub
#47 (comment).

I ran a "git pull" yesterday.

$ ./pyff --version
pyff version 0.10.0dev

Olivier Salaün

GIP RENATER
Etudes et Projets Applicatifs (EPA)
Tél : +33 2 23 23 71 27
Logo RENATER
http://www.renater.fr

[leifj]

This is now a feature-request in #62