setedit-users — General list about setedit

Re: Request: new command: cmcSearchAgainBackward

[Salvador E. T. <sal...@in...>]

Sergio N=FA=F1ez wrote:

>     Ok, here is Sergio and Ivan writing together.
>     Our personal taste about how the search should work is this (and=20
> is borrowed from UltraEdit):
>        - the user selects some text.
>        - it presses CTRL+L and searches forward for the next ocurrence =

> of that selected text.
>         OR
>        - it presses CTRL+SHIFT+L and searches backward for the=20
> ocurrencies.


Well, I don't like it. I think selecting text shouldn't affect searches.

>     The search is relative to the current cursor position, so if it is =

> backwards it searches the previous match with respect to the cursor=20
> position.
>     If the search reaches the end or the beginning of the text (if it=20
> is forward or backwards respectively) then it cycles (if it is forward =

> and reaches the end, then it searches from the beginning again).
>     Basically with this, you can do things like make shure that a=20
> given function name it is used in the places you tought without silly=20
> typos, the same for variables, or to search for a variable definition=20
> which it is on the same source.


For this you can do a search indicating "Scope: Global".

>     If the search is more complicated than this, then it would not be=20
> so practical, so we think that using first CTRL+Q+A to initialize the=20
> search isn't so practical for this type of search, instead, if the=20
> user has something selected then when pressing CTRL+L or CTRL+SHIFT+L=20
> it should start to search for that without requiring CTRL+Q+A first.=20
> So both CTRL+L or CTRL+Q+A would initialize a search, but CTRL+L is=20
> special in the sense that it can be used to repeat a search if the=20
> user has not selected a text, if it has selected text, then it=20
> initializes.
>     We hope you can understand this, otherwise we can try to explain=20
> it again, maybe in spanish language :).


I think I get the idea, but it could be implemented as a completly=20
different command. Ctrl+L is assigned to "Search Again" and this can't=20
be confused with "take the selection and search"

>     We don't understand exactly what you said about this new command.
>     Anyway, is good to know something is going to be implemented,=20
> maybe not exactly as we would like it but should be useful anyway.


The command I have in mind is just "Search again, but in the reverse=20
direction". The reeference is what you indicated in the Ctrl+Q+A dialog.

SET

--=20
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: se...@co... se...@ie...=20
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Re: Request: new command: cmcSearchAgainBackward

[<sn...@es...>]

    Ok, here is Sergio and Ivan writing together.
    Our personal taste about how the search should work is this (and is 
borrowed from UltraEdit):
       - the user selects some text.
       - it presses CTRL+L and searches forward for the next ocurrence 
of that selected text.
        OR
       - it presses CTRL+SHIFT+L and searches backward for the ocurrencies.
    The search is relative to the current cursor position, so if it is 
backwards it searches the previous match with respect to the cursor 
position.
    If the search reaches the end or the beginning of the text (if it is 
forward or backwards respectively) then it cycles (if it is forward and 
reaches the end, then it searches from the beginning again).
    Basically with this, you can do things like make shure that a given 
function name it is used in the places you tought without silly typos, 
the same for variables, or to search for a variable definition which it 
is on the same source.
    If the search is more complicated than this, then it would not be so 
practical, so we think that using first CTRL+Q+A to initialize the 
search isn't so practical for this type of search, instead, if the user 
has something selected then when pressing CTRL+L or CTRL+SHIFT+L it 
should start to search for that without requiring CTRL+Q+A first. So 
both CTRL+L or CTRL+Q+A would initialize a search, but CTRL+L is special 
in the sense that it can be used to repeat a search if the user has not 
selected a text, if it has selected text, then it initializes.
    We hope you can understand this, otherwise we can try to explain it 
again, maybe in spanish language :).
    We don't understand exactly what you said about this new command.
    Anyway, is good to know something is going to be implemented, maybe 
not exactly as we would like it but should be useful anyway.
    Thanks!!!
    Goodbye.



Salvador Eduardo Tropea wrote:

> Ivan Baldo wrote:
>
>>  Sergio and I second that wish!!! Is quite useful, something like the 
>> F3 key on UltraEdit32.
>> Even a slow and non definitive solution would do, since source code 
>> isn't that big the majority of times right? Ok, I am not suggesting 
>> following the bad practices found on other crappy software, but could 
>> an exception be done on this case given the usefulness?
>> I know I know... I am not in a good position lately to ask 
>> something... just suggesting... :)
>
>
>
> Ok, a first attempt towards it is in the current CVS code (Glaciar 
> Perito Moreno II)
> The approach is quite slow and if you try to search using a regular 
> expression that must match at the begginging/end of a line it will 
> ignore it.
> I'm quite sure it have more than one bug so please be patient and 
> report cases where it fails.
> Currently I only implemented an option in the dialog to choose 
> "backward" direction.
>
> I'm thinking about what command should be added. I don't like 
> cmcSearchAgainBackward, instead I'm thinking in something like 
> "cmcSearchAgainRev". Where "Rev" stands for "Reverse". So if you 
> created a forward search this command will be like Ctrl+L but 
> backwards. But if you created a backward search Ctrl+L will be 
> backward and this command forward. The idea is that you create a 
> search (f or b) and then Ctr+L is a "do it again" and the other is "do 
> it again, but in the reverse direction".
>
> What about it? Opinions ...
>
> SET
>


-- 
*Sergio Núñez *
Programación & Desarrollo
sn...@es... <mailto:sn...@es...>
Esquemas.com <http://www.esquemas.com/>
+ (598 2) 403 03 03
Nueva Palmira 2241
11800 Montevideo, Uruguay

Re: Request: new command: cmcSearchAgainBackward

[Salvador E. T. <sal...@in...>]

Ivan Baldo wrote:

>  Sergio and I second that wish!!! Is quite useful, something like the 
> F3 key on UltraEdit32.
> Even a slow and non definitive solution would do, since source code 
> isn't that big the majority of times right? Ok, I am not suggesting 
> following the bad practices found on other crappy software, but could 
> an exception be done on this case given the usefulness?
> I know I know... I am not in a good position lately to ask 
> something... just suggesting... :)


Ok, a first attempt towards it is in the current CVS code (Glaciar 
Perito Moreno II)
The approach is quite slow and if you try to search using a regular 
expression that must match at the begginging/end of a line it will 
ignore it.
I'm quite sure it have more than one bug so please be patient and report 
cases where it fails.
Currently I only implemented an option in the dialog to choose 
"backward" direction.

I'm thinking about what command should be added. I don't like 
cmcSearchAgainBackward, instead I'm thinking in something like 
"cmcSearchAgainRev". Where "Rev" stands for "Reverse". So if you created 
a forward search this command will be like Ctrl+L but backwards. But if 
you created a backward search Ctrl+L will be backward and this command 
forward. The idea is that you create a search (f or b) and then Ctr+L is 
a "do it again" and the other is "do it again, but in the reverse 
direction".

What about it? Opinions ...

SET

-- 
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: se...@co... se...@ie... 
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Re: Adding support for Python list of functions

[Salvador E. T. <sal...@in...>]

Grzegorz Adam Hankiewicz wrote:

>On 2004-11-29, Salvador Eduardo Tropea <sal...@in...> wrote:
>  
>
>>>What is the mode parameter to Searchxxx function?
>>>
>>>      
>>>
>>const int modeBFPrototypes=0, modeBFFunctions=1, modeBFClassSep=2;
>>modeBFPrototypes: For the list of prototypes.  modeBFFunctions:
>>For the functions list.
>>    
>>
>
>In a prototype less language I guess it's ok if I do the same for
>both of those, which is what I'm doing now, right?
>  
>

I agree.

>>modeBFClassSep: It means we want the function name to show it to
>>the user so we should show qualifiers (like class scope) *after*
>>the name i.e. Function (Class) instead of the real function name
>>(Class::Function).
>>    
>>
>
>Hmmm... yes, I was thinking about something like that. Interestingly
>in Python you can have nested function definitions, so it could
>also be used for that. However, I was thinking of more a hierarhical
>way to express it. So you could have something like:
>
>some_function(basdasd)
>some_function.little_helper_one(asdasd)
>some_function.little_helper_two(balaha)
>
>But that would be very tiresome to type, so no good use. Then I
>thought maybe using space identation like normal python code. Say:
>
>some_aaaaa()
>some_bbbb()
>some_cccc()
>  some_ccc_helper()
>  some_ccc_helper()
>some_dddd()
>
>That woult be more pleasant visually, but again, not very good to
>get at typing. Besides, I guess setedit performs its own sort at a
>higher level, so it would be even more hacky to force one or another.
>Bah, lets tell users not to use nested functions to avoid this.
>  
>

I think you should add entries like this:

little_helper_one(asdasd) [from some_function]
little_helper_two(balaha) [from some_function]

some_function(basdasd)

Then the user can do simple searchs and if the cursor is over a function 
call you can jump to the function quite fast.

>  
>
>>I know it sound crazy but the length you have to return must
>>include the EOS character. So you just need to add 1.
>>    
>>
>
>It's not weird if you work with utf8, I guess.
>
>  
>
>>I tested the code and it seems to work ok, good job.
>>    
>>
>
>Cool. I guess you are still preparing some release and this could
>get in after that?
>
>
>  
>
I already tagged the CVS and changed version to 0.5.5 ;-)
I'm waiting to have as much packages as posible before announcing the 
new release. But if you visit the web page you'll see that 0.5.4 is the 
current stable release.

SET

-- 
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: se...@co... se...@ie... 
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Re: Adding support for Python list of functions

[Grzegorz A. H. <gr...@ti...>]

On 2004-11-29, Salvador Eduardo Tropea <sal...@in...> wrote:
> >What is the mode parameter to Searchxxx function?
> > 
> const int modeBFPrototypes=0, modeBFFunctions=1, modeBFClassSep=2;
> modeBFPrototypes: For the list of prototypes.  modeBFFunctions:
> For the functions list.

In a prototype less language I guess it's ok if I do the same for
both of those, which is what I'm doing now, right?

> modeBFClassSep: It means we want the function name to show it to
> the user so we should show qualifiers (like class scope) *after*
> the name i.e. Function (Class) instead of the real function name
> (Class::Function).

Hmmm... yes, I was thinking about something like that. Interestingly
in Python you can have nested function definitions, so it could
also be used for that. However, I was thinking of more a hierarhical
way to express it. So you could have something like:

some_function(basdasd)
some_function.little_helper_one(asdasd)
some_function.little_helper_two(balaha)

But that would be very tiresome to type, so no good use. Then I
thought maybe using space identation like normal python code. Say:

some_aaaaa()
some_bbbb()
some_cccc()
  some_ccc_helper()
  some_ccc_helper()
some_dddd()

That woult be more pleasant visually, but again, not very good to
get at typing. Besides, I guess setedit performs its own sort at a
higher level, so it would be even more hacky to force one or another.
Bah, lets tell users not to use nested functions to avoid this.

> I know it sound crazy but the length you have to return must
> include the EOS character. So you just need to add 1.

It's not weird if you work with utf8, I guess.

> I tested the code and it seems to work ok, good job.

Cool. I guess you are still preparing some release and this could
get in after that?

Re: Adding support for Python list of functions

[Salvador E. T. <sal...@in...>]

Grzegorz Adam Hankiewicz wrote:

>Hello there.
>
>I'm trying to add support for Python list of functions. Please
>see the attached patch, which kind of works. Things which I have
>problems with:
>
>What is the mode parameter to Searchxxx function?
>  
>
const int modeBFPrototypes=0, modeBFFunctions=1, modeBFClassSep=2;
modeBFPrototypes: For the list of prototypes.
modeBFFunctions: For the functions list.
modeBFClassSep: It means we want the function name to show it to the 
user so we should show qualifiers (like class scope) *after* the name 
i.e. Function (Class) instead of the real function name (Class::Function).

>The callback function receives a length value. Is this the length
>of the passed string or some other length?
>

Yes, that's the length of the buffer, you shouldn't process more bytes 
than what length says.

> Is it ok passing -1 as lineEnd? I think I saw that somewhere else.
>  
>

Yes, that's ok, when you do it the code assumes the function ends just 
before the next starts.

>When I run the file as a standalone program on the attached test
>case I obtain:
>
>`show_program_usage(binary = None, exit_code = 0)' [18,-1:48]
>`mix(string_list)' [25,-1:16]
>`extract_track_number(filename)' [31,-1:30]
>`move_file(source, destination)' [40,-1:30]
>`archive_songs(index_filename, dest_dir, files)' [51,-1:46]
>`move_song(filename, dest_dir, cd_data, music_data)' [62,-1:50]
>`process_commandline(arguments, binary_name = None)' [68,-1:50]
>`main(args, binary_name = None)' [81,-1:30]
>
>Which apparently works perfectly. However, when running inside the
>editor, both the mix and show_program_usage functions have a single
>character of garbage at the end and I'm unable to jump to them,
>easily seen if I select the browse function by name or line.
>  
>

I know it sound crazy but the length you have to return must include the 
EOS character. So you just need to add 1.

I tested the code and it seems to work ok, good job.

SET

-- 
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: se...@co... se...@ie... 
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Adding support for Python list of functions

[Grzegorz A. H. <gr...@ti...>]

Hello there.

I'm trying to add support for Python list of functions. Please
see the attached patch, which kind of works. Things which I have
problems with:

What is the mode parameter to Searchxxx function?

The callback function receives a length value. Is this the length
of the passed string or some other length? Is it ok passing -1 as
lineEnd? I think I saw that somewhere else.

When I run the file as a standalone program on the attached test
case I obtain:

`show_program_usage(binary = None, exit_code = 0)' [18,-1:48]
`mix(string_list)' [25,-1:16]
`extract_track_number(filename)' [31,-1:30]
`move_file(source, destination)' [40,-1:30]
`archive_songs(index_filename, dest_dir, files)' [51,-1:46]
`move_song(filename, dest_dir, cd_data, music_data)' [62,-1:50]
`process_commandline(arguments, binary_name = None)' [68,-1:50]
`main(args, binary_name = None)' [81,-1:30]

Which apparently works perfectly. However, when running inside the
editor, both the mix and show_program_usage functions have a single
character of garbage at the end and I'm unable to jump to them,
easily seen if I select the browse function by name or line.

Re: pmacro stopped working

[Thiago F.G. A. <tf...@te...>]

> The problem is the X11 secondary clipboard (middle mouse button). When 
> available the editor keeps track of selections and informs X11 that we have a 
> selection. But the process to "copy" to the OS clipboard is stopping the 
> selection.
> I think the code that stops the selection had some sense in the old times 
> when cmcSelectOn/Off didn't exist. So I disabled it, but only when we are 
> using the "selection clipboard".
> Attached is the patch, please try it.

It works. Thank you.

Only now I noticed that this bug only occured in X.

[],
Thiago

Re: Request: new command: cmcSearchAgainBackward

[Ivan B. <ib...@ad...>]

  Sergio and I second that wish!!! Is quite useful, something like the 
F3 key on UltraEdit32.
Even a slow and non definitive solution would do, since source code 
isn't that big the majority of times right? Ok, I am not suggesting 
following the bad practices found on other crappy software, but could an 
exception be done on this case given the usefulness?
I know I know... I am not in a good position lately to ask something... 
just suggesting... :)
Love and respect everyone!!!




Salvador Eduardo Tropea escribio':

> Thiago F.G. Albuquerque wrote:
>
>> Sometimes, I miss the ability to search backwards. I'd like to have a 
>> new command:
>>
>> cmcSearchAgainBackward
>>
>> It would be like cmcSearchAgain, but it would search in the oposite 
>> direction.
>
>
> This isn't trivial, specially for regex.
> Some time ago I thinked about a solution, not perfect but quite usable.
> Is in my TODO. Remind me after the release.
>
> SET
>

-- 
Ivan Baldo - ib...@ad... - http://ibaldo.codigolibre.net/
ICQ 10215364 - Phone/FAX (598) (2) 613 3223.
Caldas 1781, Malvin, Montevideo, Uruguay, South America, planet Earth.
In favour of world peace and freedom, against U.S.A. and allies!!!
Alternatives: ib...@co... - http://go.to/ibaldo

Re: setedit/rhide recent snapshot doen't work as root

[Salvador E. T. <sal...@in...>]

Kevin Alm wrote:

>Compiled and installed setedit and rhide from yesterday's snapshot. Works fine 
>as a normal user but not as root. Not a big problem, usually use vi for 
>sysadmin stuff anyway, but this is a change from the last time I compiled 6 
>or 8 months ago. Is this an intentional change, or something else. It is 
>possible that it is due to some difference in my system config, as this new 
>install is on an alternate linux partition. Launching setedit as root results 
>in a cli blurb (setedit version, etc.) followed by an umlauted i, a 
>rectangle, a flashing underscore, and a hung computer. Control-Alt-Del reboot 
>is the only way to recover. Any ideas welcome.
>  
>
If you experiment different behavior when using root I think the problem 
is with the /dev/vcs* usage.
Are you using frame buffer? The frame buffer was always buggy and looks 
like 2.6.x kernels made it even worst.
Please give me more details about your system:
1) Linux distro.
2) Kernel version.
3) Video board, is it using a generic fb? etc.
To verify the problem is with /dev/vcs* devices try creating /root/.tvrc 
containing:

[TV]
{
[Linux]
{
 UseVCS=0
}
}

If this fixes the problem then your kernel have a nasty bug in the 
/dev/vcs* stuff.
I never saw problems when using 2.4.x, 2.2.x and 2.0.x and no frame 
buffer, but Michel Catudal (bbcat) reported very wrong colors (not a 
hang) when using Fedora Core 3 (kernel 2.6.x+frame buffer enabled).

Regards, SET

-- 
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: se...@co... se...@ie... 
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Re: pmacro stopped working

[Salvador E. T. <sal...@in...>]

Thiago F.G. Albuquerque wrote:

> For some reason, the sLisp macro 'pmacro' stopped working. I isolated 
> the problem to this snippet:
>
> (defmacro 'pmacro'
>  (progn
>   (SendCommands cmcSelectOn cmcCharLeft cmcCharLeft cmcSelectOff)
>   (ShowInStatusLine (GetSelection))
>  )
> )
>
> Maybe the the problem is in GetSelection. Or maybe when the first 
> "cmcCharLeft" is sent, it disables the selection, I don't know.

The problem is the X11 secondary clipboard (middle mouse button). When 
available the editor keeps track of selections and informs X11 that we 
have a selection. But the process to "copy" to the OS clipboard is 
stopping the selection.
I think the code that stops the selection had some sense in the old 
times when cmcSelectOn/Off didn't exist. So I disabled it, but only when 
we are using the "selection clipboard".
Attached is the patch, please try it.

Regards, SET

-- 
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: se...@co... se...@ie... 
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

pmacro stopped working

[Thiago F.G. A. <tf...@te...>]

Hi,

For some reason, the sLisp macro 'pmacro' stopped working. I isolated the 
problem to this snippet:

(defmacro 'pmacro'
  (progn
   (SendCommands cmcSelectOn cmcCharLeft cmcCharLeft cmcSelectOff)
   (ShowInStatusLine (GetSelection))
  )
)

Maybe the the problem is in GetSelection. Or maybe when the first 
"cmcCharLeft" is sent, it disables the selection, I don't know.

[],
Thiago

-- 
()  ascii ribbon campaign - against html mail 
/\                        - against microsoft attachments

setedit/rhide recent snapshot doen't work as root

[Kevin A. <kev...@sh...>]

Compiled and installed setedit and rhide from yesterday's snapshot. Works fine 
as a normal user but not as root. Not a big problem, usually use vi for 
sysadmin stuff anyway, but this is a change from the last time I compiled 6 
or 8 months ago. Is this an intentional change, or something else. It is 
possible that it is due to some difference in my system config, as this new 
install is on an alternate linux partition. Launching setedit as root results 
in a cli blurb (setedit version, etc.) followed by an umlauted i, a 
rectangle, a flashing underscore, and a hung computer. Control-Alt-Del reboot 
is the only way to recover. Any ideas welcome.

Regards,
Kevin Alm

Re: MSVC fixes for Tree view etc

[Grzegorz A. H. <gr...@ti...>]

On 2004-11-16, Arnold Wiegert <awi...@te...> wrote:
> >Commandline, built with the universal makefile or whatever SET
> >calls it.
> >
> Guess I'll have to give it a try as well; which debugger do
> you use?

The one that comes with the IDE. Pops up automatically if something
crashes.

> BTW, should I be able to debug the Linux version with DDD or do
> I need to use setedit?  I've had no luck so far with DDD and have
> never tried setedit for debugging.

No idea. I've used gdb in the past with success.

Re: MSVC fixes for Tree view etc

[Arnold W. <awi...@te...>]

Grzegorz Adam Hankiewicz wrote:

>>Do you use the command line compiler or the IDE?
>>    
>>
>
>Commandline, built with the universal makefile or whatever SET
>calls it.
>
>  
>
Guess I'll have to give it a try as well; which debugger do you use?

>>I have not tried browsing network files, but have only tried to
>>make the 'Tree' view work after starting work on UNC.
>>    
>>
>
>I thought this tree view was simply disabled, it doesn't work under
>Linux for me.
>  
>
It doesn't for me either, only under Windows Me  & 98
BTW, should I be able to debug the Linux version with DDD or do I need 
to use setedit?
I've had no luck so far with DDD and have never tried setedit for debugging.

>Anyway, if you want to try my patches I've put them at
>http://gradha.no-ip.org:8000/. I've also put a binary of the editor.
>Don't trust it much. From time to time it still does some weird
>stuff.
>
>  
>
I've downloaded it and it seems to run well enough for what I've tried 
so far - which isn't all that much yet.

Arnold

Re: MSVC fixes for Tree view in "file open" dialog

[Grzegorz A. H. <gr...@ti...>]

On 2004-11-15, Salvador Eduardo Tropea <sal...@in...> wrote:
> Well, that's the part I explained to Arnold ;-) The cl/needs.h
> definition to enable UNC code is NEEDS_UNC. At least in my
> local copy MSVC have it enabled. But it should be enabled also
> for MinGW.  The second change is in pathvalid.c, not currently
> enabled. Currently the CVS code says:
> 
> #ifdef TVComp_BCPP
> #define CheckUNC(path) \
> 
> Arnold's code enables it for MSVC and MinGW.

Ah, ok. Well, with that I'm able to browse network paths as well. So
I guess the next feature is browsing shares, then browsing computers
on a net, then integrating some HTML viewer into Setedit and claim
it cannot be split without loosing functionality, etc, etc. Clever.

On 2004-11-16, Arnold Wiegert <awi...@te...> wrote:
> Do you use the command line compiler or the IDE?

Commandline, built with the universal makefile or whatever SET
calls it.

> I have not tried browsing network files, but have only tried to
> make the 'Tree' view work after starting work on UNC.

I thought this tree view was simply disabled, it doesn't work under
Linux for me.

Anyway, if you want to try my patches I've put them at
http://gradha.no-ip.org:8000/. I've also put a binary of the editor.
Don't trust it much. From time to time it still does some weird
stuff.

Re: MSVC fixes for Tree view in "file open" dialog

[Arnold W. <awi...@te...>]

Grzegorz Adam Hankiewicz wrote:

>On 2004-11-14, Arnold Wiegert <awi...@te...> wrote:
>  
>
>>I've traced it back to tvision\compat\fexpand.c- CLY_fexpand(),
>>which does not check for UNC paths handed to it. I'm sure I can
>>come up with a hack, but I thought it would be best to get your
>>comments on it first.
>>    
>>
>
>Looks like we are on the same boat Arnold. Just recently I compiled
>the editor with msvc at work and found similar problems. I'm
>attaching the patches to setedit/tvision I've come so far to fix
>them, which are scarily more complex than yours.
>  
>
Do you use the command line compiler or the IDE?

>On the other hand, I have to say that UNC paths work ok for me if
>I pass them on the commandline as arguments, but the file dialog
>doesn't allow me to browse UNC paths because they are invalid.
>  
>
I have not tried browsing  network files, but have only tried to make 
the 'Tree' view work after starting work on UNC.

Arnold

Re: MSVC fixes for Tree view in "file open" dialog

[Salvador E. T. <sal...@in...>]

Grzegorz Adam Hankiewicz wrote:

>>P.S. For Grzegorz: Arnold enabled the UNC code used for Borland
>>compiler. This code enables the use of Windows shares
>>    
>>
>
>Indeed, I traced the open file dialog problem to CLY_PathValid,
>but that fails for me because CheckUNC is disabled for me. What
>defines should I put and where to make that code available and test
>it with my version of the patches?
>  
>
Well, that's the part I explained to Arnold ;-)
The cl/needs.h definition to enable UNC code is NEEDS_UNC. At least in 
my local copy MSVC have it enabled. But it should be enabled also for MinGW.
The second change is in pathvalid.c, not currently enabled. Currently 
the CVS code says:

#ifdef TVComp_BCPP
#define CheckUNC(path) \

Arnold's code enables it for MSVC and MinGW.
A third change that I can remmember is the definition of the CLY_IsUNC 
macro in compatlayer.h.
I think that this is currently defined as a working function for MSVC 
and BC++, it should also include MinGW.
I don't remmeber the other changes and some of them will collide with 
your changes because Arnold's copy still using \ as dir separator.

So you need a diff -u output from Arnold's copy.

SET

-- 
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: se...@co... se...@ie... 
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Re: MSVC fixes for Tree view in "file open" dialog

[Grzegorz A. H. <gr...@ti...>]

On 2004-11-14, Arnold Wiegert <awi...@te...> wrote:
> I've traced it back to tvision\compat\fexpand.c- CLY_fexpand(),
> which does not check for UNC paths handed to it. I'm sure I can
> come up with a hack, but I thought it would be best to get your
> comments on it first.

Looks like we are on the same boat Arnold. Just recently I compiled
the editor with msvc at work and found similar problems. I'm
attaching the patches to setedit/tvision I've come so far to fix
them, which are scarily more complex than yours.

On the other hand, I have to say that UNC paths work ok for me if
I pass them on the commandline as arguments, but the file dialog
doesn't allow me to browse UNC paths because they are invalid.

I have to say that there are still missing things in my patch. For
example, when Setedit calls getenv and retrieves a path with
backslashes, that should be converted to forward ones, but I still
haven't written the correct CLY_ function for it.

> P.S. For Grzegorz: Arnold enabled the UNC code used for Borland
> compiler. This code enables the use of Windows shares

Indeed, I traced the open file dialog problem to CLY_PathValid,
but that fails for me because CheckUNC is disabled for me. What
defines should I put and where to make that code available and test
it with my version of the patches?

Re: MEPIS Linux 2.6.7 & rhtivision 2.0.3

[Arnold W. <awi...@te...>]

Salvador Eduardo Tropea wrote:

> Arnold Wiegert wrote:
>
>> Hi Salvador,
>>
>> One of my hard drives hiccupped and I had to install a new linux ( 
>> along with all of Winxx) on my machine and tried SimplyMepis; I'm now 
>> running the 2.6.7 kernel and just downloaded the latest snapshots 
>> from sourceforge.
>>
> kotori:~/Persistencia$ dpkg -S term.h
> libncurses5-dev: /usr/include/term.h
>
> This header is part of ncurses.
>
That fixed it;
Guess I need to become more familiar with configure

>> PS: Mepis was the simplest install of a debian based system I've ever 
>> run into; it even recognized all of my hardware without any help on 
>> my part. Impressive!
>
>
>
> Didn't know about Mepis, but the main reason I have to use Debian is 
> its stability and coherence. For the Debian stable that's far better 
> than most Linux distros and at the same time you have a remarkable 
> system to help upgrading and fixing security problems. I think the 
> combo is much better than any other Linux distro.
> The problem with most distros derived from Debian is that they doesn't 
> use the stable base and hence you lose the most important features.
>
I can appreciate your concerns. Although, for me, this distro seems the 
best so far.

Trying to do it all with the stable version and by hand, I had lots of 
problems. My video cards was unsupported and needed a lengthy trial and 
error - mostly error - period.
Similarly the network interface seemed suspect at times. About half the 
time the system complained about time-out after booting, although I 
never did conclusively find any specific problem or solution.

None of these problems - nor any other - have shown up with MEPIS; it 
simply loads and runs; the only manual 'work' has been to install 
packages which were not part of the default.

I'm quite happy with it - finally something that seems even easier and 
more reliable than Windows :-)

Then again, perhaps I just got lucky this time ;-)

Arnold

Re: MSVC fixes for Tree view in "file open" dialog

[Salvador E. T. <sal...@in...>]

Arnold Wiegert wrote:

> Hello, Salvador
>
> finally got some time to look at more of the Windows problems I was 
> having.
>
> The latest version had problems with the "File Open" dialog; the 
> 'Tree' view seemed to be ignored.
>
> To fix it, I finally got to implement some of your recommendations 
> from some time ago,:-)


I'll be analyzing it after the release. I hope it will be soon.
Can you compile *all* the changes in a diff generated with -u ?

I'll have to merge things for the UNC with things from Grzegorz.

Regards, SET

-- 
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: se...@co... se...@ie... 
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Re: MEPIS Linux 2.6.7 & rhtivision 2.0.3

[Salvador E. T. <sal...@in...>]

Arnold Wiegert wrote:

> Hi Salvador,
>
> One of my hard drives hiccupped and I had to install a new linux ( 
> along with all of Winxx) on my machine and tried SimplyMepis; I'm now 
> running the 2.6.7 kernel and just downloaded the latest snapshots from 
> sourceforge.
>
> After running ./configure, I tired compiling and got the following error
>
> root@4[tvision]# make
> make -C makes -f librhtv.mkf
> make[1]: Entering directory `/usr/aww/pkg/set/tvision/makes'
> gcc  -I../include        -O2 -pipe  -c ../classes/linux/linuxdis.cc -o 
> obj/linuxdis.o
> ../classes/linux/linuxdis.cc:33:18: term.h: No such file or directory
> make[1]: *** [obj/linuxdis.o] Error 1
> make[1]: Leaving directory `/usr/aww/pkg/set/tvision/makes'
> make: *** [static-lib] Error 2
>
>
> There is not term.h file on my system.


kotori:~/Persistencia$ dpkg -S term.h
libncurses5-dev: /usr/include/term.h

This header is part of ncurses.

> Arnold
>
> PS: Mepis was the simplest install of a debian based system I've ever 
> run into; it even recognized all of my hardware without any help on my 
> part. Impressive!


Didn't know about Mepis, but the main reason I have to use Debian is its 
stability and coherence. 
For the Debian stable that's far better than most Linux distros and at 
the same time you have a remarkable system to help upgrading and fixing 
security problems. I think the combo is much better than any other Linux 
distro.
The problem with most distros derived from Debian is that they doesn't 
use the stable base and hence you lose the most important features.

> There are a ferw nasties with the UNC path on the Windows version of 
> 0.5.4. build 1145 but I haven't had the time to track them down


Ok.

SET

P.S. For Grzegorz: Arnold enabled the UNC code used for Borland 
compiler. This code enables the use of Windows shares 
(//machine/share/file).

-- 
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: se...@co... se...@ie... 
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Re: Request: new command: cmcSearchAgainBackward

[Salvador E. T. <sal...@in...>]

Thiago F.G. Albuquerque wrote:

> Sometimes, I miss the ability to search backwards. I'd like to have a 
> new command:
>
> cmcSearchAgainBackward
>
> It would be like cmcSearchAgain, but it would search in the oposite 
> direction.

This isn't trivial, specially for regex.
Some time ago I thinked about a solution, not perfect but quite usable.
Is in my TODO. Remind me after the release.

SET

-- 
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: se...@co... se...@ie... 
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Request: new command: cmcSearchAgainBackward

[Thiago F.G. A. <tf...@te...>]

Sometimes, I miss the ability to search backwards. I'd like to have a new 
command:

cmcSearchAgainBackward

It would be like cmcSearchAgain, but it would search in the oposite 
direction.

[],
Thiago

-- 
()  ascii ribbon campaign - against html mail 
/\                        - against microsoft attachments

MSVC fixes for Tree view in "file open" dialog

[Arnold W. <awi...@te...>]

Hello, Salvador

finally got some time to look at more of the Windows problems I was having.

The latest version had problems with the "File Open" dialog; the 'Tree' 
view seemed to be ignored.

To fix it, I finally got to implement some of your recommendations from 
some time ago,:-)

 >From Sept 24 - for compatlayer.h
 >I think the real problem is this:
 >
 >/* SET: Not sure why Vadim wanted it. Perhaps to make the port more 
usable
 >   for users accustomed to back slashes, but then the library isn't 
uniform
 >   across platforms */
 >#define DIRSEPARATOR '\\'
 >#define DIRSEPARATOR_ "\\"

The files I had to modify were
======================
compatlayer.h
39c39
< #include <configtv.h>
---
 > #include <tv/configtv.h>
451d450
<
1357,1361c1356,1357
< //define DIRSEPARATOR '\\'        // AWW
< // #define DIRSEPARATOR_ "\\"    // AWW
<  #define DIRSEPARATOR '/'        // AWW
<  #define DIRSEPARATOR_ "/"        // AWW
<
---
 >  #define DIRSEPARATOR '\\'
 >  #define DIRSEPARATOR_ "\\"
=========================================
diff between my tvision\classes\tdirlist.cpp and yours
22c22
< #include <configtv.h>         ##### ignore, MSVC does not seem to like 
realtive paths
---
 > #include <tv/configtv.h>
503,504c503
< //    #if defined(TVCompf_djgpp) || defined(TVComp_BCPP)    // AWW
<     #ifndef TVOS_UNIX                                        // AWW
---
 >     #if defined(TVCompf_djgpp) || defined(TVComp_BCPP)


=========================================
diff between my and your setedit\settvuti\fileopen.cc
125,145c125,128
< #define AWW
< #if defined( AWW )
<  char *pt;
< 
<  strcpy( buf2, startDir );
<  for ( pt = buf2; *pt; pt++ )
<  {
<      if ( *pt == '\\' )
<      {
<          memmove( pt+2, pt+1, strlen(pt+1)+1 );
<          *(pt+1) = '\\';
<          pt++;
<      }
<  }
<  if (!getcwd(buf,PATH_MAX) || (startDir && !chdir(buf2)))
<      return NULL;
<
< #else
<  if (!getcwd(buf,PATH_MAX) || (startDir && !chdir(startDir)))
<     return NULL;
< #endif
---
 >
 >  if (!getcwd(buf,PATH_MAX) || (startDir && !chdir(startDir)))
 >     return NULL;
 >

without this change, chdir will always cause an immediate return; it has 
to be double forward slashes; backslashes don't work.

I'm not sure how or where you want to implement this code; but so far it 
works for me as  is.

============================
The only problem I have left (of those I know about ;-) ) is that when I 
open a file on a different machine by double-clicking on it - i.e. via 
file association - with setedit,  the resulting path is not interpreted 
correctly and it fails to open the file at all.

Doing it on the local machine, produces the correct file in standard MS 
format with back slashes. Doing it across the network produces a file 
name such as:
@:\Silver\d\pkg\c\li\s\linter.c
instead of simply
\\silver\d\pkg\c\li\s\linter.c
Similarly, calling Setedit in a DOS box with the same name has the same 
effect, except the file name displayed at the top of the setedit screen 
like this:
c:\silver\d\pkg\c\li\c\linter.c
In both these cases, setedit assumes a new file name and does not open 
the specified file.

I've traced it back to tvision\compat\fexpand.c- CLY_fexpand(), which 
does not check for UNC paths handed to it. I'm sure I can come up with a 
hack, but I thought it would be best to get your comments on it first.

Arnold