[ANN] Gjallar release "Eldrimner" 0.1

Hi folks!

We are pleased to announce the first public availability of Gjallar (see
http://swiki.krampe.se/gjallar) as release "Eldrimner", modestly
numbered 0.1, typically for daredevil developers only. The license for
Gjallar is the new BSD license.

Note that this is merely a decent checkpoint during development, Gjallar
is not yet ready for anything else than experimental use and of course -
development.

But it does *work* by some definition and we are about to start small
test deployments at our customer. But we haven't fully tested it, there
are glaring things missing and probably also things missing that we do
not know about. :)

We are trying hard to make Gjallar easily accessible for development.
One thing is that we have prepared a development image ready for
download. We are also writing down different procedures and describing
the design. We have not yet written a detailed tested tutorial/user
guide, but it is coming.

If Gjallar seems interesting and you wish to help, make addons or even
base a product around the codebase - feel free to contact us to get on
the team and join the mailinglist - we are currently using the setools
list.

For more info (including screenshots) check the homepage:

        http://swiki.krampe.se/gjallar

regards, Göran & Magnus

PS. If I messed something up when preparing the release, sorry. It was
late. :) Feel free to email privately or post to the setools list.

PPS. Yes, a public Gjallar server for handling Gjallar bugs and as a
demo will be coming too.

[hidden email] wrote:
> We are pleased to announce the first public availability of Gjallar (see
> http://swiki.krampe.se/gjallar) as release "Eldrimner", modestly
> numbered 0.1, typically for daredevil developers only. The license for
> Gjallar is the new BSD license.

Excellent work. A great example that newcomers to Squeak/Seaside
(and Magma) can use. Congrats and thanks to you and the sponsors
who've open-sourced it.

I downloaded the zip file, ran some doIt's from the workspace,
and then the login appeared in the browser. (There were a few
missteps along the way, but nothing serious).

There were some odd pauses (hour glass cursor), but a little
patience and they cleared. I had to cmd-. to regain control at
one point - I think it was trying to send mail, and timing out.
I followed the method that created the sample data and eventually
figured out some userid/pwd to use. Unfortunately, that's about
all the time I have to look at it for the moment.

Hi!

Yanni Chiu <[hidden email]> wrote:
> [hidden email] wrote:
> > We are pleased to announce the first public availability of Gjallar (see
> > http://swiki.krampe.se/gjallar) as release "Eldrimner", modestly
> > numbered 0.1, typically for daredevil developers only. The license for
> > Gjallar is the new BSD license.
>
> Excellent work. A great example that newcomers to Squeak/Seaside
> (and Magma) can use. Congrats and thanks to you and the sponsors
> who've open-sourced it.

Thanks. Yes, I also think that this app can be used as a larger "sample"
app which you could copy stuff from.
We have done quite a bit of generally useful things, like a syslog-ish
Log mechanism, a Scheduler a la cron, a bunch of session classes (with
and without Seaside) with Magma integration, a connection pool (mostly
copied from Kilauea, even though I have probably patched it into
oblivion now), a large scale use of the Command pattern blablabla...

> I downloaded the zip file, ran some doIt's from the workspace,
> and then the login appeared in the browser. (There were a few
> missteps along the way, but nothing serious).

Please tell and I can tweak these things to get it as painless as
possible, even if its details. :)

> There were some odd pauses (hour glass cursor), but a little

When initializing the db there is a lot of stuff going on.

> patience and they cleared. I had to cmd-. to regain control at
> one point - I think it was trying to send mail, and timing out.

Ok. Possibly. Or rather, it probably tried popping. The default settings
for POP3 is bogus of course but by default all Q2Services are activated
so it will try to pop. I forgot to deactivate most of those services in
the default image. The services available are Q2Service subclasses and
each has a class comment. I think you would want to deactivate all to
start with using:

        Q2Scheduler deactivateAllServices  (IIRC)

> I followed the method that created the sample data and eventually
> figured out some userid/pwd to use. Unfortunately, that's about
> all the time I have to look at it for the moment.

No problem :). The initial accounts etc are also described at the Swiki:

        http://swiki.krampe.se/gjallar/12

...should perhaps have pointed to the Swiki more forcefully in that
Doits window. :)

regards, Göran

Hi,

Good job  : http://swiki.krampe.se/gjallar, it's very usefull for me :-)

1/
About architecture I've tried to understand this part on the
requirements document, but it's for me a difficult point ;-).

"Sripting" seems to be very usefull but can you use it to move case for
example?

2/
What's the main idea to make a distinction between Process and
ProcessPrototype ?

thanks,

Flo

Hi Florent!

florent trolat <[hidden email]> wrote:
> Hi,
>
> Good job  : http://swiki.krampe.se/gjallar, it's very usefull for me :-)

I will try to get even more info up there.
 
> 1/
> About architecture I've tried to understand this part on the
> requirements document, but it's for me a difficult point ;-).
>
> "Sripting" seems to be very usefull but can you use it to move case for
> example?

Today there is no Scripting mechanism implemented but we had a meeting
just today where a few things popped up that would be best dealt with
using simple scripting, so I will probably add it soon.

The idea is to be able to tie Smalltalk code easily into both forms and
as "handlers" of events. One example is to fill in a calculated value in
one field when editing another. Another example is more complex
notification triggers, or more advanced integration needs.

> 2/
> What's the main idea to make a distinction between Process and
> ProcessPrototype ?

An organization like Micronic (or any one) has a LOT of different issue
processes - some of them are clearly visible and known, others are
"hidden" in the way people work - like a group of people sending an
Excel document around gathering information, or a group of people
sharing an email account in order to have a common "inbox" for issues
etc.

So the system should easily support setting up new Processes that are
"pre configured". The best way to do that is to be able to clone an
existing Process which has been configured. So there is no class
ProcessPrototype - a prototype of a Process is just a Process that is
not kept in the list of "real" Processes and is only used by cloning it
to get a new such Process.

The same technique exists in the system for creating Cases (by cloning
an existing prototype case) and Users (or at least that was the idea,
not implemented yet IIRC).

> thanks,
>
> Flo

regards, Göran

[hidden email] wrote:
>>I downloaded the zip file, ran some doIt's from the workspace,
>>and then the login appeared in the browser. (There were a few
>>missteps along the way, but nothing serious).
>
> Please tell and I can tweak these things to get it as painless as
> possible, even if its details. :)

1) When I started the image, it was unclear whether the server
was already started. So I tried to browse anyway, and got no
connection.

2) Looking at the various doIt's and the method in the open
class browser, I chose something that said it would start the
server. It did, but I got a stack trace in the web browser
(something about a connection - maybe it said Magma, but that
was my guess). When I'd earlier browsed the files that were
unzipped, it didn't seem to have any database files. So, the
next step was to create the database. I chose to create the
database with sample data, waiting patiently for the hourglass
to stop fading in/out. (A message that database creation had
completed would be useful).

3) Bingo! I got a login, but I didn't know a usrid/pwd to use.
I tried the "forgot your password" link. Then I noticed that
the image seemed frozen. Although the image was not responsive,
the website was still functioning. After about two minutes of
occasional cmd-. attempts, I eventually got a debugger. Then I
browsed the code to find a usrid/pwd, and eventually found them.

Suggestion: a single readme window (the method in the open class
browser had code that looked useful as doIt steps, so it was a
distraction). The readme would just have steps 1, 2, 3, etc.,
which would get you a login page, and tell you a usrid/pwd to
use. A swiki link might be good, but it could be out of sync
with any particular image. A few steps in the actual image might
be better (and it turns out that it was just a few simple steps).

Hi Yanni!

Yanni Chiu <[hidden email]> wrote:
Ok, yes - you can save the image with everything running - but I wanted
to distribute a "clean" image.
But I will make sure people knows that.
 
> 2) Looking at the various doIt's and the method in the open
> class browser, I chose something that said it would start the
> server. It did, but I got a stack trace in the web browser
> (something about a connection - maybe it said Magma, but that
> was my guess).

Right, it should say clearer that you need to initializeDatabase first.

> When I'd earlier browsed the files that were
> unzipped, it didn't seem to have any database files.

Exactly, it will be created etc from scratch if missing.

> So, the
> next step was to create the database. I chose to create the
> database with sample data, waiting patiently for the hourglass
> to stop fading in/out. (A message that database creation had
> completed would be useful).

Good idea since this will always be run from the image anyway.

> 3) Bingo! I got a login, but I didn't know a usrid/pwd to use.

Ehh, ok, I will make a clearer pointer to the Swiki for this kind of
info.
And/or add it to the texts in the image.

> I tried the "forgot your password" link. Then I noticed that
> the image seemed frozen. Although the image was not responsive,

Ok, will look into that. Hmmm.

I agree - good feedback, thanks. I will fix it.

regards, Göran

Looks Great!

Some little problems.

Logging off and then backarrow puts it in an undefined state. You can
still back through your prior pages but clicking on links give a
walkback. Some continuation shouldn't be there after a logoff.

Backarrow also gives the "next state" list box of the issue the wrong
list of next states.

Would be nice to have a single list of all files/folders needed/used,
in the docs w/ a description of their purpose.

Might be nice to have a builtin wiki.

Can we have a field to add a comment during /any/ change posted to the issue?

I'm looking forward to see how this developes!

Thank you and thanks to your sponsor.

Cheers,
Darius

Hi!

"Darius Clarke" <[hidden email]> wrote:
> Looks Great!

Thanks!

> Some little problems.
>
> Logging off and then backarrow puts it in an undefined state. You can
> still back through your prior pages but clicking on links give a
> walkback. Some continuation shouldn't be there after a logoff.
>
> Backarrow also gives the "next state" list box of the issue the wrong
> list of next states.

Right, we haven't bothered to even check such things. I suppose we need
to add a bunch of Seaside backtracking thingies etc. Feel free to help
out :).

> Would be nice to have a single list of all files/folders needed/used,
> in the docs w/ a description of their purpose.

Will be added ASAP.
 
> Might be nice to have a builtin wiki.

There actually is a small implementation of a wiki inside. But more work
is needed on it, we intend to try to do that soon - it is connected to
the help system and documentation approach.

> Can we have a field to add a comment during /any/ change posted to the issue?

Like a "commit comment"? Mmm, good idea. Either that would be an extra
field that is available close to the "Save" button when editing (or any
of the other actions that can be made below the case tab book) and it
would end up visible in the history view, and/or it would generate a
"Note" like the other notes, but associated with the txn made.
 
> I'm looking forward to see how this developes!

Yeah! :)
 
> Thank you and thanks to your sponsor.

We both just thinks this is the best path to follow for us too. OSS when
it works the best. :)

> Cheers,
> Darius

regards, Göran