Kudos and nits

First things first - nice job, guys.  I was very pleased to find
that the professional image came up with many of the tools I use
regularly (particularly the refactoring browser, the SUnit
framework, and the SUnit browser) already loaded and ready.  I
appreciate this.

I've got a few minor problems to report:

   1.  Open a package browser and right-click on a changed package.
       Click on the Source Control menu entry.  Note that the
       'Check Out Changes' menu item is enabled.  Click on the
       'Check Out Changes' menu item.  You'll get a walkback
       titled
          PackageSelector does not understand checkOutChanges

   2.  When searching for text in methods using the Method - Browse -
       Containing Text... menu item on a system browser the cursor
       never changes to a wait cursor (hourglass) - it remains an
       arrow during the search.  Not sure if this is related to the
       problems reported earlier regarding cursor handling.  This
       worked OK under 4.01.3.

   3.  Method - Browse - Containing Text... doesn't find anything
       if the search text begins with a '#'.  For example, try
       searching for '#changed'.  Nothing is found, even though
       '#changed' appears in the text of Package>>isChanged.  Again,
       this worked as expected under 4.01.3.  (This problem makes it
       tedious to find all places where a given event is triggered).

And a suggestion:

Now that the SUnit framework is shipped as part of Dolphin, could
Dolphin be changed so that 'unit tests' is suggested as a candidate
category for all methods starting with 'test'?  Perhaps you could
add the following to MethodCategorizationPolicy>>suggestionsFor:

        "Does it begin in 'test'"
        ('test*' match: selector) ifTrue: [
                suggestions add: (categoryManager name: 'unit tests')].

This would simplify the categorization of test methods in TestCase
subclasses.

"Bob Jarvis" <[hidden email]> wrote in message
news:[hidden email]...
Sorry about that. There are probably other similar cases that arise because
we haven't shipped our 'Source Repository' package with the image (its
always been an internal thing). I've recorded this as defect #929.
>
>    2.  When searching for text in methods using the Method - Browse -
>        Containing Text... menu item on a system browser the cursor
>        never changes to a wait cursor (hourglass) - it remains an
>        arrow during the search.  Not sure if this is related to the
>        problems reported earlier regarding cursor handling.  This
>        worked OK under 4.01.3.

The beta releases of Dolphin XP ran all commands as "modal", disabling the
originating view while they ran. This meant that a wait cursor would be
displayed even should a dialog be popped up, which usually results in the
wait cursor set by the command execution being cleared. In consequence
explicit wait cursor management was removed in certain places (such as the
full text search). However it proved to be undesirable to make commands
"modal" in general, since one doesn't always want the view disabled,
therefore this was changed to be an attribute of individual commands. The
text search should have been marked as modal, but has not been. I've
recorded this as defect #930.

>
>    3.  Method - Browse - Containing Text... doesn't find anything
>        if the search text begins with a '#'.  For example, try
>        searching for '#changed'.  Nothing is found, even though
>        '#changed' appears in the text of Package>>isChanged.  Again,
>        this worked as expected under 4.01.3.  (This problem makes it
>        tedious to find all places where a given event is triggered).

The full-text search now supports wildcards. # has traditionally been a
wildcard character in Smalltalk, matching a single character. Thus if you
enter #blah
then only methods with source code 5 characters longs ending in 'blah' will
match. Searching without wildcards is equivalent to *<text to find>*.

Full-text search is useful (particularly with wildcards), but there is a
much faster and more accurate way to search for references to arbitrary,
which is to use the prompted Browse/References To... This can be invoked
from the system folder main menu, or by pressing F12 (or using the context
menu command) in any workspace. Searching for references (or definitions in
fact) can also be done with wildcards.

Yes, I have did sometimes think that myself during the development of
Dolphin XP, but obviously not often enough to make me want to put it in!
I've recorded it as an enhancement request (#931).

Regards

Blair

> The full-text search now supports wildcards. # has traditionally been a
> wildcard character in Smalltalk, matching a single character. Thus if you
> enter #blah then only methods with source code 5 characters longs ending
> in 'blah' will match. Searching without wildcards is equivalent to
> *<text to find>*.

<puzzled look> I guess I don't see the utility of the '#' wildcard, or
of adding Yet Another Wildcarding System (YAWS).  <sigh>  I'll add
this to the list of Things I Don't Understand (along with my wife, my
kids, my co-workers, my boss, and fish), and I'll look forward to this
being documented in the new Edjakayshun Sentr Fur Duhmees.  :-)

> Full-text search is useful (particularly with wildcards), but there is a
> much faster and more accurate way to search for references to arbitrary,
> which is to use the prompted Browse/References To... This can be invoked
> from the system folder main menu, or by pressing F12 (or using the context
> menu command) in any workspace. Searching for references (or definitions in
> fact) can also be done with wildcards.

This worked perfectly.  Thanks for the help!

Bob Jarvis

"Bob Jarvis" <[hidden email]> wrote in message
news:[hidden email]...
> > The full-text search now supports wildcards. # has traditionally been a
> > wildcard character in Smalltalk, matching a single character. Thus if
you
> > enter #blah then only methods with source code 5 characters longs ending
> > in 'blah' will match. Searching without wildcards is equivalent to
> > *<text to find>*.
>
> <puzzled look> I guess I don't see the utility of the '#' wildcard, or
> of adding Yet Another Wildcarding System (YAWS).  <sigh>  I'll add
> this to the list of Things I Don't Understand (along with my wife, my
> kids, my co-workers, my boss, and fish), and I'll look forward to this
> being documented in the new Edjakayshun Sentr Fur Duhmees.  :-)

Like I said, it has _traditionally_ been used in Smalltalk's String
matching, so in fact I wouldn't be surprised if it didn't actually predate
the more familiar wildcard character to which you refer (perhaps a Smalltalk
historian could chip in with some background here). Full text search is
implemented on top of the standard string matching (i.e. String>>match:), so
it picks up that behaviour along the way.

Regards

Blair