Mail Archives: djgpp/1998/01/05/09:13:54
"S. M. Halloran" <mitch AT duzen DOT com DOT tr> wrote:
Some of these things must be replied by Robert, I'll put IMHO before any reply
that applies better to Robert.
> (1) [MF] In middle of window movement/resize, it is necessary to
> press 'Enter' to leave the window-movement/resize before any other
> input activity, namely mouse use, can be employed. With Turbo C-like
> editors, a mouse input action can cause a change a window to become
> active and leave the window move/resize state.
IMHO that's a very small thing and don't know if Robert will invest time on it.
Normally users use keyboard or mouse to resize, not both.
> (2) [B] Likely a 'gcc' error: in linking phase, the message window
> reports for duplicated external references the following (an
> example):
> <begin example>
> Creating: rasse.exe
> Error: ../obj/chem.o: In function `LinkHNum':
> chem.c(95) Error: multiple definition of `LinkHNum'
> o:checkcyc.c(263) Error: first defined here
> <end example>
> Note that error is in last line: clicking or hitting error on
> selected line in message which contains 'chem.c' properly opens new
> window with source 'chem.c'. Clicking line containing 'o:checkcyc.c'
> opens an empty window at position 1:1 with no source. The problem is
> the 'o:' part.
IMHO is an error in the "errors parser" of RHIDE, but can't crash your system
;-) (For the [B])
> (3) [B] Editor error: In some cases, the editor loses track of the
> cursor; this may result from complicated movments of cursor
> repositioning via arrow keys (re-positioning by mouse?), but will
> attempt to pay more careful track of what input events occur prior to
> 'lost cursor'. Not necessary to re-start rhIDE or close window under
> fault: believe that moving a complete 'windowful' up or down (use of
> PG UP or PG DN) key restores synchrony between cursor and
> character-entry position.
I think I know what's that. I can bet you have "Optimal Fill" option ON. If
that's true that's an known problem in my editor v0.3.6. When you use Optimal
Fill the editor fills the gaps with tabs+spaces but the routine that
calculates the needed tabs/spaces fails for gaps smallers than a tab.
Reducing the tab size you'll reduce the probability to hit the bug, dissabling
the Optimal Fill you won't hit the bug but the gaps will be filled with spaces.
> (4) [MF] Two items are missing from the Window menu of RHide. One
> is the ability to call up the Watch window (key combination is Alt-W
> + 'W' in Borland's system) and the Message window (key combo is
> Alt-W + 'M'). In fact, a closed Message window apparently cannot be
> recovered, although a closed Watch window can (via Ctrl-F7). The key
> combination for the Project window summons, currently Alt-W + 'J'
> seems to be at variance with Borland's key combination of Alt-W + 'P'
> (this is a recommended change to be consistent with the Borland
> interface). The only way to summon the Message Window is to do a
> build/make/compile at present, not a desirable situation.
Is Robert's stuff, I talked it with you before, so no additions here.
> (5) [MF] Borland IDE has a much better presentation of structures and
> arrays during debug sessions through the data 'Inspector' system.
> This is a separate window showing members and their values (for
> structures) or the numbered array element and its value (for arrays).
> The Watch window currently allows a line-oriented presentation which
> is difficult to manage.
Yes that's the thing that I really miss from BC. But implementing that is a
really problem. To understand the problem you must know that RHIDE uses the GNU
debugger to debug (gdb). GDB is inside the IDE and the IDE talks with gdb in a
similar way that a user does. This mechanism helps the user because the gdb
interface is hard to learn and provides a very good debugging tool without
writing a new debuger, but have limitations.
Actually is possible to implement some thing similar to an inspector but it
will have the following problems:
1) The code will be dependent of gdb and the debug info used, so new versions
of gdb or changes in the debug info can produce problems.
2) Will be limited to the gdb features.
But the main problem is the time. I don't have time to code it and I don't know
if Robert have (I doubt).
If you look in the gdb output for a class you'll see you can create an
inspector window parsing this long answer but isn't so simple.
> (6) [MF] Cannot alter presentation style of value in Watch window as
> can be done in Borland IDE. Borland allows alternative
> formatting/presenation for ALL defined symbols in the Watch window.
> This is in addition to being able to cast a defined symbol. For
> example, integer variables can be expressed as decimals or
> hexadecimals (Borland Turbo C++ v.3.0 did not allow octal expression,
> but that would be a useful feature).
Are you sure? I think gdb have some modifiers, don't know if they are available
from the evaluate window, Robert must ask it.
> (7) [B] Running on a 486DX (33 MHz) with 4 MB RAM--and very slowly I
> might add--with VM=35 MB and PM=1 MB and only the minimal drivers
> (after TSRs loaded, free conventional RAM = 612K or so), RHide
> sometimes exits abruptly after a big build and then a step (Shift-F7
> or F7) into a debug session. This remains to be seen on a "real"
> system (a Pentium @ 100 MHz with 16 MB RAM and ~100 MB free disk
> space).
No idea why in the Pentium.
> (8) [MF] The List of Opened Windows (key combo Alt-'0' or Alt-W + 'L'
> apparently only features opened windows. Other files (closed
> windows?) present do not represent an ordered list of recently opened
> windows, as is the case in the Borland IDE. This list of recently
> opened windows is a very handy feature.
Don't understand your request, are you asking for an special sort in the closed
windows list?
> (9) [MF] The Project window currently sorts all source files in
> alphabetical order, which is at variance with the more
> preferable Borland IDE system of sorting sources according to the
> position in which they are added with respect to the selection line.
But I find it MUCH more usefull. My project have tons of sources and I can make
incremental searchs like that: ALT+1, type 2 or 3 letters, press ENTER and I
get the right source. That's easier for sorted projects. We discussed in the
group why a user can want a non-sorted project and the conclusion was that only
a few (<1%) needs it and can be solved changing the names of the sources.
Robert is planing to add an option for non-sorted projects.
> Moreover, the Borland IDE presented a summary of the lines and bytes
> of the code and data sections compiled (perhaps gcc can be coaxed
> into reporting this, although this is admittedly a low priority
> feature).
I don't know about any switch to get this information.
SET
------------------------------------ 0 --------------------------------
Visit my home page: http://set-soft.home.ml.org/
or
http://www.geocities.com/SiliconValley/Vista/6552/
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Alternative e-mail: set-sot AT usa DOT net - ICQ: 2951574
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA
TE: +(541) 759 0013
- Raw text -