[kepler-dev] Re: [SDM-SPA] Re: Kepler Trial Release

Fri Mar 5 21:59:31 PST 2004

Matt:

I agree: we will need more formal and systematic testing (e.g., we can 
adopt the Ptolemy way of doing that, or invent our own).

For now I still like to encourage Zhengang and Ilkay to push forward
with the alpha level "internal release" of version 0 (that should be
understatement enough ;-)

For example in SPA we have a long time user of earlier versions of the 
tool (including hand-crafted Perl scripts) and we want to provide him
with a tool that works for him; at the same time we can learn what's
missing for the end user etc.

I guess over the next few months in coordination with all of Kepler we
will be able to provide the more systematic testing approach that you
suggest. For example for the SEEK meeting in May we should have an
improved infrastructure in place. But there is an even earlier
deadline for us, the SDM PI meeting later this month. It would be good
if, for example Terence or Arie would be able to install and
demonstrate a pre-release version then (btw: I can't make it to that
meeting).

Bertram

>>>>> "MJ" == Matt Jones <jones at nceas.ucsb.edu> writes:
MJ> 
MJ> David et al.,
MJ> I agree that the kepler installer should be seamless for the end user. 
MJ> Its the whole point of having an installer.
MJ> 
MJ> As I've mentioned before, I think the InstallAnywhere system produces 
MJ> quite usable installers for Windows and Macintosh and Linux.  We haven't 
MJ> experienced any significant difficulties using this system.  I would 
MJ> suggest that evaluating other installers than InstallShield would be 
MJ> well-worth our while given recent experiences with InstallShield.
MJ> 
MJ> I also think that much more formal and systematic testing of kepler and 
MJ> the installers will circumvent the types of problems that you've 
MJ> encountered.  That we don't have these formal test procedures in place 
MJ> is exactly why I suggested we aren't really ready for anything other 
MJ> than an alpha release with very limited distribution.
MJ> 
MJ> At a minimum, whoever creates the installer should test it on versions 
MJ> of Windows (98, 2K, XP, 2003) with and without a previous install and 
MJ> with all supported JVM version, and versions of MacOS once we've worked 
MJ> out the mouse issues and decide to support MacOS.  Developers personal 
MJ> workstations tend to be bad test environemnts because they typically are 
MJ> highly configured and don't represent the typical user experience. 
MJ> Better to try it on non-development machines like lab machines where the 
MJ> tester does not have admin rights.
MJ> 
MJ> Cheers,
MJ> Matt
MJ> 
MJ> David Buttler wrote:
MJ> 
>> Okay, maybe I am being dense, but I am still having problems.
>> 
>> I uninstalled Ptolemy. I removed kepler and every SWE directory I 
>> could.  I emptied my env. variables of reference to ptolemy or kepler
>> 
>> I went to the Ptolemy site and downloaded the monster ptolemy 
>> distribution that includes its own java runtime (the one that says that 
>> this is the distribution that everyone should get)
>> 
>> I ran the ptolemy installer; no problems detected. Everything installed 
>> well.
>> I ran the kepler installer.  It gave me a set of options (modify, 
>> repair, remove).  The default selected option was "remove"
>> Two things seem wrong to me at this point:
>> 1) I am installing for the first time.  I don't want to modify, repair, 
>> or remove.  I want to install.
>> 2) Of all of the options that are available why is "remove" the default 
>> selected option?
>> 
>> Well, okay, perhaps modify is the option I want.  I click this.  It 
>> lists SWE as the component to install, claiming that it will absorb 0 K 
>> of my disk space, out of 11 GB free. (minor quibble).  I click on OK, 
>> and it starts to install, but then bails after about 5 seconds with the 
>> error "1203"
>> I don't know what this means.  I am a reasonably savvy computer user, 
>> but any numeric error code is meaningless to me.  If we distribute 
>> software that produces this error, we will loose a lot of potential users.
>> 
>> I recently tried installing a commercial package that gave me similar 
>> useless error message.  I guessed that it was because I do not have 
>> administrative rights (and I don't have write permissions on anywhere 
>> except my own "My Docs" directory, and a special folder (c:\usr) I had 
>> created so that I could work on compiling windows stuff without having 
>> to fight directory names that contain spaces).  In the previous case, 
>> this seemed to be the problem; I had an administrator walk over to my 
>> office, log in, install my software, and walk back to his office.  
>> Elapsed time, 3 days.  I don't want to have to do that with Kepler; not 
>> because I am too lazy, but because I know that users here have about as 
>> much control over their desktops as I have over mine, and it is not 
>> acceptable to force an administrator to install this software.  If 
>> Ptolemy doesn't need an administrator to install, then we shouldn't either.
>> 
>> Honestly, I am not trying to create extra work for Zhengang, or whoever 
>> eventually creates the install script.  I would be happy to look at the 
>> InstallSheild script to help identify what can be changed to improve the 
>> install process.  Several years ago, back at OGI, I had the 
>> responsibility to use InstallSheild to create a distribution of software 
>> we were working on then, so I know more than nothing about it.  Perhaps, 
>> if we understood the process of InstallSheild, we (either me or our new 
>> developer Xiaowen) can create an InstallAnywhere script.  [which would 
>> be useful for unix/linux, macs, as well as windows]
>> 
>> Dave
>> 
>> Ilkay Altintas wrote:
>> 
>>> David,
>>> 
>>> Please uninstall Ptolemy and uninstall Kepler in this order.
>>> And then reinstall Ptolemy and then reinstall Kepler.
>>> This solves some of the installation problems related to
>>> file updates. Might help...
>>> 
>>> Ilkay
>>> 
>>> -----Original Message-----
>>> From: sdm-dev-admin at sdsc.edu [mailto:sdm-dev-admin at sdsc.edu]On Behalf Of
>>> David Buttler
>>> Sent: Friday, March 05, 2004 4:36 PM
>>> To: Zhengang Cheng
>>> Cc: sdm-dev at sdsc.edu
>>> Subject: [SDM-SPA] Re: Kepler Trial Release
>>> 
>>> 
>>> Hi Zhengang,
>>> 
>>> I am having the same trouble with the Kepler release as before.  The
>>> installation looks like it gets half-way through and then it stops with
>>> an unintelligible error.  What can I do to help track down this
>>> problem?  Do you have an InstallSheild script that I could look at to
>>> see if I can get a better handle on what is going on during the install?
>>> 
>>> Thanks,
>>> Dave
>>> 
>>> Zhengang Cheng wrote:
>>> 
>>> 
>>> 
>>>> Deal All,
>>>> 
>>>> With Ilkay's great help, I are releasing the Alpha 0 release of Kepler.
>>>> The URL is at http://sdm2.csc.ncsu.edu/SWE/kepler.exe.
>>>> This release contains the work from SPA, SEEK, GEON etc.
>>>> 
>>>> We know there are errors (eg. in geon, icon in desktop) , and we will 
>>>> work
>>>> to improve next week. The main purpose of this release is for 
>>>> everyone have
>>>> a general view of Kepler later release. Your suggestion for 
>>>> improvement is
>>>> certainly welcome.
>>>> 
>>>> Thanks,
>>>> 
>>>> Zhengang
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
MJ> 
MJ> 
MJ> _______________________________________________
MJ> kepler-dev mailing list
MJ> kepler-dev at ecoinformatics.org
MJ> http://www.ecoinformatics.org/mailman/listinfo/kepler-dev