Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Fri, Apr 21, 2017 at 6:51 AM, john lunzer  wrote:

> Ahh, I'm sorry to have suggested urwid. I totally forgot about the Linux
> only limitation.
>

​Not a problem.  Getting up to speed with winpdb/rpdb2 makes everything
worthwhile.

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Friday, April 21, 2017 at 6:24:11 AM UTC-5, Edward K. Ream wrote:

In short, urwid doesn't seem to work on Windows, but npyscreen runs 
> anywhere and can be debugged on Linux with rpdb2 and winpdb.  I'd say this 
> is a good enough resolution to the problem of what curses widgets to use.
>

At last I see. There is a much easier shortcut to using rpdb2 or winpdb.  
Just invoke them without arguments:

rpdb2 launchLeo.py --gui=curses

or

   winpdb launchLeo.py --gui=curses

This launches rpdb2 or winpdb in the original console, and creates a 
*second* console to run the launchLeo.py. The server in the second console 
connects automatically with the client/debugger in the original console.  
So everything "just works".

This is more than good enough. No need for commercial alternatives.

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[john lunzer]

Blasphemy! Haha, jk. I too probably would have recommended pycharm if I 
were going to go with anything "pro". As much as I try not to give 
commercial software kudos they've got a good piece of software.

It's a shame the debugging experience on Windows isn't more fluid. I'm not 
sure how long I'd survive with Python on Linux without pudb. At least it 
wouldn't be nearly as fun. Another reason to lament urwid not working on 
Windows.

On Friday, April 21, 2017 at 8:14:26 AM UTC-4, Kent Tenney wrote:
>
> or https://www.jetbrains.com/pycharm/
>
>
> On Fri, Apr 21, 2017 at 7:11 AM, Kent Tenney  > wrote:
>
>> It might be worth biting the bullet, buying a commercial IDE like
>> https://wingware.com/
>>
>>
>> On Fri, Apr 21, 2017 at 6:51 AM, john lunzer > > wrote:
>>
>>> Ahh, I'm sorry to have suggested urwid. I totally forgot about the Linux 
>>> only limitation. 
>>>
>>>
>>> On Friday, April 21, 2017 at 7:24:11 AM UTC-4, Edward K. Ream wrote:

 On Friday, April 21, 2017 at 5:56:29 AM UTC-5, Edward K. Ream wrote:
>
>
> Ok.  Misleading 'help attach' and error message confused me for awhile.
>

 Sorry, I mis-remembered what I had just done!

 In the second console, first do:

 password 

 Then do:

 attach

 to get a list of pid's. Then do:

 attach 

 to actually attach to the first console.

 This *does* work, but alas rpdb2 doesn't print *anything* useful after 
 single-stepping, so I'm always having to type l (list).  So using the 
 winpdb gui version is probably best after all.

 I've just tried to use rpdb2 in both consoles on Windows.  There is a 
 problem connecting to the newly-created server on localhost.  Something 
 about a timeout.  Not sure if this is a programming problem (fctrl?) or a 
 firewall problem.  However, there is no great need to do debugging on 
 Windows.

 In short, urwid doesn't seem to work on Windows, but npyscreen runs 
 anywhere and can be debugged on Linux with rpdb2 and winpdb.  I'd say this 
 is a good enough resolution to the problem of what curses widgets to use.

 Edward

 -- 
>>> You received this message because you are subscribed to the Google 
>>> Groups "leo-editor" group.
>>> To unsubscribe from this group and stop receiving emails from it, send 
>>> an email to leo-editor+...@googlegroups.com .
>>> To post to this group, send email to leo-e...@googlegroups.com 
>>> .
>>> Visit this group at https://groups.google.com/group/leo-editor.
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
>>
>

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Kent Tenney]

or https://www.jetbrains.com/pycharm/


On Fri, Apr 21, 2017 at 7:11 AM, Kent Tenney  wrote:

> It might be worth biting the bullet, buying a commercial IDE like
> https://wingware.com/
>
>
> On Fri, Apr 21, 2017 at 6:51 AM, john lunzer  wrote:
>
>> Ahh, I'm sorry to have suggested urwid. I totally forgot about the Linux
>> only limitation.
>>
>>
>> On Friday, April 21, 2017 at 7:24:11 AM UTC-4, Edward K. Ream wrote:
>>>
>>> On Friday, April 21, 2017 at 5:56:29 AM UTC-5, Edward K. Ream wrote:


 Ok.  Misleading 'help attach' and error message confused me for awhile.

>>>
>>> Sorry, I mis-remembered what I had just done!
>>>
>>> In the second console, first do:
>>>
>>> password 
>>>
>>> Then do:
>>>
>>> attach
>>>
>>> to get a list of pid's. Then do:
>>>
>>> attach 
>>>
>>> to actually attach to the first console.
>>>
>>> This *does* work, but alas rpdb2 doesn't print *anything* useful after
>>> single-stepping, so I'm always having to type l (list).  So using the
>>> winpdb gui version is probably best after all.
>>>
>>> I've just tried to use rpdb2 in both consoles on Windows.  There is a
>>> problem connecting to the newly-created server on localhost.  Something
>>> about a timeout.  Not sure if this is a programming problem (fctrl?) or a
>>> firewall problem.  However, there is no great need to do debugging on
>>> Windows.
>>>
>>> In short, urwid doesn't seem to work on Windows, but npyscreen runs
>>> anywhere and can be debugged on Linux with rpdb2 and winpdb.  I'd say this
>>> is a good enough resolution to the problem of what curses widgets to use.
>>>
>>> Edward
>>>
>>> --
>> You received this message because you are subscribed to the Google Groups
>> "leo-editor" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to leo-editor+unsubscr...@googlegroups.com.
>> To post to this group, send email to leo-editor@googlegroups.com.
>> Visit this group at https://groups.google.com/group/leo-editor.
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Kent Tenney]

It might be worth biting the bullet, buying a commercial IDE like
https://wingware.com/


On Fri, Apr 21, 2017 at 6:51 AM, john lunzer  wrote:

> Ahh, I'm sorry to have suggested urwid. I totally forgot about the Linux
> only limitation.
>
>
> On Friday, April 21, 2017 at 7:24:11 AM UTC-4, Edward K. Ream wrote:
>>
>> On Friday, April 21, 2017 at 5:56:29 AM UTC-5, Edward K. Ream wrote:
>>>
>>>
>>> Ok.  Misleading 'help attach' and error message confused me for awhile.
>>>
>>
>> Sorry, I mis-remembered what I had just done!
>>
>> In the second console, first do:
>>
>> password 
>>
>> Then do:
>>
>> attach
>>
>> to get a list of pid's. Then do:
>>
>> attach 
>>
>> to actually attach to the first console.
>>
>> This *does* work, but alas rpdb2 doesn't print *anything* useful after
>> single-stepping, so I'm always having to type l (list).  So using the
>> winpdb gui version is probably best after all.
>>
>> I've just tried to use rpdb2 in both consoles on Windows.  There is a
>> problem connecting to the newly-created server on localhost.  Something
>> about a timeout.  Not sure if this is a programming problem (fctrl?) or a
>> firewall problem.  However, there is no great need to do debugging on
>> Windows.
>>
>> In short, urwid doesn't seem to work on Windows, but npyscreen runs
>> anywhere and can be debugged on Linux with rpdb2 and winpdb.  I'd say this
>> is a good enough resolution to the problem of what curses widgets to use.
>>
>> Edward
>>
>> --
> You received this message because you are subscribed to the Google Groups
> "leo-editor" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to leo-editor+unsubscr...@googlegroups.com.
> To post to this group, send email to leo-editor@googlegroups.com.
> Visit this group at https://groups.google.com/group/leo-editor.
> For more options, visit https://groups.google.com/d/optout.
>

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[john lunzer]

Ahh, I'm sorry to have suggested urwid. I totally forgot about the Linux 
only limitation. 

On Friday, April 21, 2017 at 7:24:11 AM UTC-4, Edward K. Ream wrote:
>
> On Friday, April 21, 2017 at 5:56:29 AM UTC-5, Edward K. Ream wrote:
>>
>>
>> Ok.  Misleading 'help attach' and error message confused me for awhile.
>>
>
> Sorry, I mis-remembered what I had just done!
>
> In the second console, first do:
>
> password 
>
> Then do:
>
> attach
>
> to get a list of pid's. Then do:
>
> attach 
>
> to actually attach to the first console.
>
> This *does* work, but alas rpdb2 doesn't print *anything* useful after 
> single-stepping, so I'm always having to type l (list).  So using the 
> winpdb gui version is probably best after all.
>
> I've just tried to use rpdb2 in both consoles on Windows.  There is a 
> problem connecting to the newly-created server on localhost.  Something 
> about a timeout.  Not sure if this is a programming problem (fctrl?) or a 
> firewall problem.  However, there is no great need to do debugging on 
> Windows.
>
> In short, urwid doesn't seem to work on Windows, but npyscreen runs 
> anywhere and can be debugged on Linux with rpdb2 and winpdb.  I'd say this 
> is a good enough resolution to the problem of what curses widgets to use.
>
> Edward
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Friday, April 21, 2017 at 5:56:29 AM UTC-5, Edward K. Ream wrote:
>
>
> Ok.  Misleading 'help attach' and error message confused me for awhile.
>

Sorry, I mis-remembered what I had just done!

In the second console, first do:

password 

Then do:

attach

to get a list of pid's. Then do:

attach 

to actually attach to the first console.

This *does* work, but alas rpdb2 doesn't print *anything* useful after 
single-stepping, so I'm always having to type l (list).  So using the 
winpdb gui version is probably best after all.

I've just tried to use rpdb2 in both consoles on Windows.  There is a 
problem connecting to the newly-created server on localhost.  Something 
about a timeout.  Not sure if this is a programming problem (fctrl?) or a 
firewall problem.  However, there is no great need to do debugging on 
Windows.

In short, urwid doesn't seem to work on Windows, but npyscreen runs 
anywhere and can be debugged on Linux with rpdb2 and winpdb.  I'd say this 
is a good enough resolution to the problem of what curses widgets to use.

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Friday, April 21, 2017 at 5:47:56 AM UTC-5, Edward K. Ream wrote:

This is pretty cool.  I actually would prefer not to use the winpdb gui.  
> At present, I haven't figured out how I might use rpdb2 in the *second* 
> console.
>

Ok.  Misleading 'help attach' and error message confused me for awhile.  At 
first, the proper form of attach is 'attach ', where the password 
is the one given in the *first* console.

attach  yields a list process id's.  Then attach  attaches 
to the first console.  Now you can use the *second* rpdb2 pretty much like 
pdb to debug the script in the first console.  It's excellent.

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Friday, April 21, 2017 at 4:19:43 AM UTC-5, Edward K. Ream wrote:

So I would like to try urwid on Windows, but neither pudb nor winpdb seem 
> to work on Windows, due to problems with fctrl on Windows. 
>

Alright.  Some real progress. winpdb works well on Leo as follows:

1.  In a console, do:

rpdb2 -d -r launchLeo.py --gui=curses

This creates a server process which controls execution of the script.  Do 
*not* specify the python executable!

2. In a *separate* console, do:

winpdb

3. Choose File:Attach in winpdb. This attaches winpdb to the server created 
by rpdb2.

Now you are ready to execute pdb-like commands from the command line in 
winpdb.  I just did g (go), and the *first* console executed as usual.

Actually, the first time I ran the go command there was a "window too small 
exception" which winpdb caught.  After detaching, the traceback appeared in 
the *first* (rpdb2) window.

This is pretty cool.  I actually would prefer not to use the winpdb gui.  
At present, I haven't figured out how I might use rpdb2 in the *second* 
console.

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Friday, April 21, 2017 at 4:19:43 AM UTC-5, Edward K. Ream wrote:

> For now, the way forward will be to duplicate the npyscreen code in urwid 
on Windows.

The following code fails with both Python 2 and 3 on Windows:

import urwid
txt = urwid.Text(u"Hello World")
fill = urwid.Filler(txt, 'top')
loop = urwid.MainLoop(fill)
loop.run() # On Windows, fails differently on 2 and 3

The failures on Windows involve either missing termios or fctrl, depending 
on python version.

So urwid is not going to get the job done on Windows.  A stack-overflow item 
 recommends using cygwin on 
windows. That's conceivable, but for now I need a break from all of this. 
We can discuss options at the sprint...

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Friday, April 21, 2017 at 3:58:49 AM UTC-5, Edward K. Ream wrote:

Yesterday was, mostly, a long digression into portability, installation and 
> debugging issues. Not fun. These diversions are a great way to kill 
> enthusiasm.
>

Importing urwid works in windows with 2/3.  On Linux, urwid is installed in 
home/local.../site-packages (python 2) which is on the path, and on 
/usr/local.../dist-packages (python 3), which isn't on my path.  This kind 
of thing is wearing me out.

Also, on Linux, pip is screaming at me to upgrade to 9.0.1, even after I 
have done so. 

So I would like to try urwid on Windows, but neither pudb nor winpdb seem 
to work on Windows, due to problems with fctrl on Windows.  Can *nobody* 
simulate fctrl (pipes) on Windows?? Some days I am just fed up with it 
all...

For now, the way forward will be to duplicate the npyscreen code in urwid 
on Windows.  I don't need a debugger for that.  Then we'll see about *dashed 
*Linux install paths...

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Thursday, April 20, 2017 at 6:38:33 AM UTC-5, Edward K. Ream wrote:
>
> On Thursday, April 20, 2017 at 6:21:05 AM UTC-5, Edward K. Ream wrote:
>>
>>
>> Rev e1e4e48 completes part A using an npyscreen.NPSApp class and 
>> associated MultiLineEditableBoxed widget. It looks like npyscreen has what 
>> we need, including a tree widget.
>>
>
> One more improvement: the various __getattr__ methods now use both g.trace 
> and g.es, so that their output is retained in the curses log widget after 
> the original console's contents is overwritten.
>
> On to part B!
>

Yesterday was, mostly, a long digression into portability, installation and 
debugging issues. Not fun. These diversions are a great way to kill 
enthusiasm.

I'll be investigating the portability (windows/linux and python 2/3) and 
usability of urwid next.  The "extra" test between python 2/3 roughly 
doubles my work. Alas, there is no way to avoid it.  Python 2 suffers about 
as many problems as 3.

One excellent (unexpected!) development was that defining just a few new 
ivars in the curses-related classes sufficed to eliminate all the 
__getattr__ methods!  I discovered this mostly by accident. Strange details 
omitted.

In hindsight, another way of getting to the present state might have been 
to base classes on the various "Null" classes in leoPy.leo#Code-->Gui base 
classes. But the __getattr__ traces have a great virtue.  They showed 
exactly *what *attributes Leo's startup code accessed and *where* the the 
startup code the accesses happened. Null classes could not have shown that.

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Thursday, April 20, 2017 at 6:09:19 PM UTC-5, Terry Brown wrote:

In linux at least there are plenty of ways of separating the trace 
> output from the screen output.  


Yes, I know that. On Linux, but not windows, the console is restored after 
curses exits, so any traces are available later, just as with logging or 
printing to a file.

But I also need to single-step through code.  g.pdb uses the console, so 
it's out. winpdb runs on Linux, so I'll come up to speed with that.

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Terry Brown]

On Thu, 20 Apr 2017 17:24:19 -0500
"Edward K. Ream"  wrote:

> On Thu, Apr 20, 2017 at 11:31 AM, Kent Tenney 
> wrote:
> 
> > possibly logging would help?
> > https://pypi.python.org/pypi/devpy
> >
> 
> ​I doubt it. The solution, imo, is to use pudb (linux) or winpdb
> (windows). I got sidetracked with other issues.  I'll attempt one or
> the other working now...

In linux at least there are plenty of ways of separating the trace
output from the screen output.  As Kent points out, you could use
logging to log to a file, and then to tail -f on the file in a
different terminal, separating tty display and logging completely.
Maybe you don't even need logging, just points g.* outputs at a file
rather than sys.stderr would do it.

Cheers -Terry

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Thu, Apr 20, 2017 at 11:31 AM, Kent Tenney  wrote:

> possibly logging would help?
> https://pypi.python.org/pypi/devpy
>

​I doubt it. The solution, imo, is to use pudb (linux) or winpdb (windows).
I got sidetracked with other issues.  I'll attempt one or the other working
now...

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Kent Tenney]

possibly logging would help?
https://pypi.python.org/pypi/devpy

provides logging with one line of code, maybe the useful message gets
logged before the terminal is lost ...

On Thu, Apr 20, 2017 at 10:38 AM, Edward K. Ream 
wrote:

> On Thu, Apr 20, 2017 at 8:12 AM, john lunzer  wrote:
>
> I realize you're doing this as an exercise but if you're getting
>> frustrated with npyscreen (I also think the docs are a little sparse) there
>> is always urwid, which I think has much better documentation. It's also got
>> all the widgets necessary for Leo.
>>
>
> ​Thanks for the reminder. Now would be a good time to investigate urwid.
>
> I'm having trouble getting control of npyscreen keystrokes in a useful
> way. The biggest challenge is debugging. Both prints (sys.stdout.write) and
> g.pdb scroll the npyscreen stuff off the console!
>
> Edward
>
> --
> You received this message because you are subscribed to the Google Groups
> "leo-editor" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to leo-editor+unsubscr...@googlegroups.com.
> To post to this group, send email to leo-editor@googlegroups.com.
> Visit this group at https://groups.google.com/group/leo-editor.
> For more options, visit https://groups.google.com/d/optout.
>

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Thu, Apr 20, 2017 at 8:12 AM, john lunzer  wrote:

I realize you're doing this as an exercise but if you're getting frustrated
> with npyscreen (I also think the docs are a little sparse) there is always
> urwid, which I think has much better documentation. It's also got all the
> widgets necessary for Leo.
>

​Thanks for the reminder. Now would be a good time to investigate urwid.

I'm having trouble getting control of npyscreen keystrokes in a useful way.
The biggest challenge is debugging. Both prints (sys.stdout.write) and
g.pdb scroll the npyscreen stuff off the console!

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[john lunzer]

I realize you're doing this as an exercise but if you're getting frustrated 
with npyscreen (I also think the docs are a little sparse) there is always 
urwid, which I think has much better documentation. It's also got all the 
widgets necessary for Leo.

Also, again, a good example of a full screen application which has a layout 
somewhat similar to what you'd expect in Leo is *pudb*. It's got a tree 
widget which lists variables and a code widget which shows source code. 
It's also got a list widget that lists the frame stack.

I've mentioned it before but there is also a full fledged code editor name 
xo (exofrills) which is also written in urwid.

My plan, when I magically found time, was to study both pudb and xo, 
because I think between the two there is enough there to put together a 
full fledged Leo curses GUI.

On Thursday, April 20, 2017 at 8:46:27 AM UTC-4, Edward K. Ream wrote:
>
> On Thursday, April 20, 2017 at 6:21:05 AM UTC-5, Edward K. Ream wrote:
>
> Remember, to run the app, use the --gui=curses command-line option. To 
>> quit the app, click on the ok button in the lower right corner, then hit 
>> return.
>>
>
> Also, when not editing, tab moves between widgets, and shift-tab moves in 
> the opposite direction.
>
> The npyscreen docs are weak.  Reading the code (or trial and error) is 
> more helpful.
>
> Edward
>

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Thursday, April 20, 2017 at 6:21:05 AM UTC-5, Edward K. Ream wrote:

Remember, to run the app, use the --gui=curses command-line option. To quit 
> the app, click on the ok button in the lower right corner, then hit return.
>

Also, when not editing, tab moves between widgets, and shift-tab moves in 
the opposite direction.

The npyscreen docs are weak.  Reading the code (or trial and error) is more 
helpful.

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Thursday, April 20, 2017 at 6:21:05 AM UTC-5, Edward K. Ream wrote:
>
>
> Rev e1e4e48 completes part A using an npyscreen.NPSApp class and 
> associated MultiLineEditableBoxed widget. It looks like npyscreen has what 
> we need, including a tree widget.
>

One more improvement: the various __getattr__ methods now use both g.trace 
and g.es, so that their output is retained in the curses log widget after 
the original console's contents is overwritten.

On to part B!

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.

Re: ENB: Curses prototype: phase 2 design and process

[Edward K. Ream]

On Thursday, April 20, 2017 at 3:26:31 AM UTC-5, Edward K. Ream wrote:

2. The cursesGui ctor set self.consoleOnly = True, which tells g.trace to 
> use print immediately for traces instead of queuing up traces for Leo's log 
> window.  Today's work will start with making things work when 
> g.app.gui.consoleOnly is False. This will queue traces for later.
>

Done at 243740. Notes:

1. Leo does *not* queue g.trace messages. Leo only queues log messages.
2. It was useful *not* to define a __getattr__ method in the CursesLog 
class.  This ensured quick tracebacks.
3. The CursesLog class is *not* a subclass of the LeoLog class, at least 
for now. Only two methods were copied from LeoLog.
4. CG.runMainLoop just calls sys.exit().  It is only a test of the 
CursesLog class.

So now we have gone as far as we can without a curses log widget.  There 
are two parts left of this phase of the prototype:

Part A.  Implement a curses log widget and re-enable curses in 
CG.runMainLoop.
Part B.  Call k.masterKeyHandler to handle keystrokes.

Each part will like take an hour or three.

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to leo-editor+unsubscr...@googlegroups.com.
To post to this group, send email to leo-editor@googlegroups.com.
Visit this group at https://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/d/optout.