[lokisetup] Problem with scripts not adding to installed_bytes
megastep at megastep.org
Tue Jun 29 23:30:31 EDT 2004
OK, I committed patches to address these issues, mostly from Derek's
work. The documentation is also updated to reflect the changes in the
For the scripts returning odd error codes, I had never observed that
behavior, and I use lots of scripts in my installers. So I'm wondering
exactly what may be going on for you, the run_script() function is
fairly simple. Though from looking at the code it could fail if your
process was terminated by a signal.
I'm hesitant about adding error messages for failed scripts too,
especially since copy.c is mostly shielded from the UI code (not that
we couldn't do it anyway). Maybe just log a debug message that can be
traced with setup -v 0 ...
On Jun 29, 2004, at 8:51 AM, Derek R. Gaston wrote:
> Ah, now I see - the size property is different for a script than it is
> for an option - I don't think that's any good. With both of my
> updates (including https://bugzilla.icculus.org/show_bug.cgi?id=1828 )
> I think the script tag should now handle the same size format as
> Should we change the documentation?
> I understand the problem with the rc - but I don't understand why it's
> giving incorrect output. I even have all of my scripts "exit 0" at
> the end - and they are still returning 1 - so I can only conclude that
> something is broken. I took a look at the code that actually calls
> the script - and I don't think I'm qualified to make any changes in
> that code.
> Maybe if someone else out there wants to give it a whirl and see if
> they can get it to return 0 out of ANY script and then tell me how to
> do it we can just revert that part - or like you said if rc!=0 maybe
> throw an installation error.
> Rob Crittenden wrote:
>> On Tue, 29 Jun 2004, Derek R. Gaston wrote:
>>> Seriously: sz comes across as "10M" in my case - it just reads the
>>> "size" property of the "script" tag.
>> Right, the size property of the script tag is supposed to be in
>> bytes, not
>> in shorthand. I can't control what is in the tag. It is good to use
>> the size_node() function in any case, I didn't see that function when
>> added this.
>> I think it is wrong to disregard the return value. I didn't have any
>> on my system with a bad return. One might argue that rc != 0 should
>> an installation error and not continue at all. How else can the script
>> programmer return a useful error message?
megastep at megastep.org
More information about the Lokisetup