CONFIGURE permission shouldn't allow the type of the job to be changed.
That's more of CREATE+DELETE.
In any case, the code doesn't correctly handling submitting config.xml
for a different type.
After talking to Jesse, he's OK with me bringing it back to public so
long as we don't allow other programmatic dependencies to it.
The intention of leaving them mutable is to allow admins to play with
this in the groovy script during the initialization and at runtime.
Groovy currently ignores the private access modifier anyway, but that is
considered as a bug in the upstream
(https://jira.codehaus.org/browse/GROOVY-3010)
If Jenkins URL is set to https, force the secure flag. Also force the
cookie to be HTTP only, which mitigates the damage that XSS can cause.
See https://www.owasp.org/index.php/SecureFlag
Don't let UsernameNotFoundException vs BadCredentialsException
difference to be seen by the caller, for that tells whether the user
exists or not.
But to assist trouble-shooting, do report that error to the server. UUID
helps the user finds the information in the log file
Ubuntu (at least as of 12.04) has the default umask 022, which made some
users nervous. Quoting its /etc/login.defs below, which explains its
historical origin:
UMASK is the default umask value for pam_umask and is used by
useradd and newusers to set the mode of the new home directories.
022 is the "historical" value in Debian for UMASK
027, or even 077, could be considered better for privacy
There is no One True Answer here : each sysadmin must make up his/her
mind.
It does seem to me that a bit more restrictive default is sensible,
so this change introduces /etc/default/jenkins parameter that sets the
default umask to 027 to prevent "others" from seeing files.
Not that keys and other sensitive files are protected anyway, so it is
not the case that the privacy of Jenkins data files have been vulnerable
prior to this change.
ZeroClipboard 1.3.5 is rather incompatible with 1.1.7, and various API changes were needed.
- setText() call doesn't work until the DOM is populated, which is at some unknown time AFAICT.
installing it via the datarequested event avoids this problem.
- constructor now demands the element to attach to, and it's unclear if relative positioning is working or not.
- "display: inline-block" is needed for ZeroClipboard to correctly compute the height of the element
java.lang.NullPointerException: null
at hudson.model.AbstractBuild.setWorkspace(AbstractBuild.java:296)
at hudson.model.ProjectTest.testCheckout(ProjectTest.java:423)
For some reason IE10 craches when document.getElementsByName is called and there is an element with a name attribute but no value. In this case a button had a name attribute set with an empty string as the value.
The Run constructors taking a timestamp must not be deprecated.
And the root directory must be created in RunMap.put, not the constructor,
in case the number (and hence directory name) is changed after the constructor completes.