Invalid 'relevant' conditions bringing down ODK Collect

Development
#1

I'm seeing that if a condition is malformed it can either not allow a
new form to be started, (e.g. relevant="selected(/data/foo, 1)" raises
an exception with a message of "java.lang.Double", which is not very
helpful) or crash the app entirely (e.g. relevant="lkasjdsd d>
alskfae12" crashes the works).

Shouldn't a poorly formed condition just be ignored (that is, default
to true)? I assume this would be as easy as wrapping an appropriate
piece of code in a try/catch and gobbling any errors arising from
xpath evaluation...

Thoughts?

#2

i'm assuming the crash is in javarosa-core, right? i'm sure everyone
would appreciate a patch...

··· On Thu, Sep 1, 2011 at 20:00, Thomas Smyth wrote: > I'm seeing that if a condition is malformed it can either not allow a > new form to be started, (e.g. relevant="selected(/data/foo, 1)" raises > an exception with a message of "java.lang.Double", which is not very > helpful) or crash the app entirely (e.g. relevant="lkasjdsd d> > alskfae12" crashes the works). > > Shouldn't a poorly formed condition just be ignored (that is, default > to true)? I assume this would be as easy as wrapping an appropriate > piece of code in a try/catch and gobbling any errors arising from > xpath evaluation... > > Thoughts? >
#3

Can you file a bug and attach the crash log? We've tried to make ODK catch
and report all the errors output from javarosa, but I have noticed some
conditions that cause javarosa to crash, which in turn causes ODK to crash.

··· On Thu, Sep 1, 2011 at 10:42 AM, Yaw Anokwa wrote:

i'm assuming the crash is in javarosa-core, right? i'm sure everyone
would appreciate a patch...

On Thu, Sep 1, 2011 at 20:00, Thomas Smyth thomas.smyth@gatech.edu wrote:

I'm seeing that if a condition is malformed it can either not allow a
new form to be started, (e.g. relevant="selected(/data/foo, 1)" raises
an exception with a message of "java.lang.Double", which is not very
helpful) or crash the app entirely (e.g. relevant="lkasjdsd d>
alskfae12" crashes the works).

Shouldn't a poorly formed condition just be ignored (that is, default
to true)? I assume this would be as easy as wrapping an appropriate
piece of code in a try/catch and gobbling any errors arising from
xpath evaluation...

Thoughts?

#4

Ok I'll give it a shot when I get a minute.

··· On 1 September 2011 13:50, Carl Hartung wrote: > Can you file a bug and attach the crash log? We've tried to make ODK catch > and report all the errors output from javarosa, but I have noticed some > conditions that cause javarosa to crash, which in turn causes ODK to crash. > > > On Thu, Sep 1, 2011 at 10:42 AM, Yaw Anokwa wrote: >> >> i'm assuming the crash is in javarosa-core, right? i'm sure everyone >> would appreciate a patch... >> >> On Thu, Sep 1, 2011 at 20:00, Thomas Smyth wrote: >> > I'm seeing that if a condition is malformed it can either not allow a >> > new form to be started, (e.g. relevant="selected(/data/foo, 1)" raises >> > an exception with a message of "java.lang.Double", which is not very >> > helpful) or crash the app entirely (e.g. relevant="lkasjdsd d> >> > alskfae12" crashes the works). >> > >> > Shouldn't a poorly formed condition just be ignored (that is, default >> > to true)? I assume this would be as easy as wrapping an appropriate >> > piece of code in a try/catch and gobbling any errors arising from >> > xpath evaluation... >> > >> > Thoughts? >> > > >