Re: [lazarus] TRunInThread - Progress, but new problem
On Sun, 5 Mar 2006, A.J. Venter wrote: On Sunday 05 March 2006 08:43, A.J. Venter wrote: Well since my last mail on this a few weeks ago, I have made quite a bit of progress in my attempts to create a component that will allow me to call a tprocess in a thread while retrieving the output live. The structure I have no compiles, and can be placed on a form - it is all but right. The only trouble is - when I call the run method, it throws an access violation, the stack trace places it at line 76 (the line where I call FProcessThread.resume), I did a check and the first line of the execute method does not ever get executed. Mmmm, I think I know now why this is happening, though now I have even LESS of an idea how to prevent it. I instantiated TProcessThread directly in a test program (obviously this left me with no events or such but I was just trying to see if it worked) when I call TProcessThread.Create the program dies with error 232 complaining that there isn't a thread driver. I checked the lpr, it calls cthreads. I checked lazarus.pp - that too calls cthreads, the only thing I can imagine is that I must somehow call cthreads in olpack, putting it in olpack.lpk doesn't work - lazarus recreates the file on compile/save apparently, putting it in runinthread.pas doesn't work (probably because it's not early enough in the uses clause) - leaving me unable to determine where I SHOULD add it - or is this the IDE telling me that what I am trying to do is simply not possible ? You should add it in the program file (.lpr) of your test program. As the _first_ unit (possibly after cmem, if you use that too). Michael. _ To unsubscribe: mail [EMAIL PROTECTED] with unsubscribe as the Subject archives at http://www.lazarus.freepascal.org/mailarchives
Re: [lazarus] TRunInThread - Progress, but new problem
I checked the lpr, it calls cthreads. I checked lazarus.pp - that too calls cthreads, the only thing I can imagine is that I must somehow call cthreads in olpack, putting it in olpack.lpk doesn't work - lazarus recreates the file on compile/save apparently, putting it in runinthread.pas doesn't work (probably because it's not early enough in the uses clause) - leaving me unable to determine where I SHOULD add it - or is this the IDE telling me that what I am trying to do is simply not possible ? You should add it in the program file (.lpr) of your test program. As the _first_ unit (possibly after cmem, if you use that too). It IS the first unit in the lpr :( Could it be that it somehow just doesn't get compiled intoto the component ? Has anybody ever DONE a threaded component ? A.J. -- there's nothing as inspirational for a hacker as a cat obscuring a bug by sitting in front of the monitor - Boudewijn Rempt A.J. Venter Chief Software Architect OpenLab International www.getopenlab.com www.silentcoder.co.za +27 82 726 5103 _ To unsubscribe: mail [EMAIL PROTECTED] with unsubscribe as the Subject archives at http://www.lazarus.freepascal.org/mailarchives
[lazarus] TRunInThread - Progress, but new problem
Well since my last mail on this a few weeks ago, I have made quite a bit of
progress in my attempts to create a component that will allow me to call a
tprocess in a thread while retrieving the output live.
The structure I have no compiles, and can be placed on a form - it is all but
right. The only trouble is - when I call the run method, it throws an access
violation, the stack trace places it at line 76 (the line where I call
FProcessThread.resume), I did a check and the first line of the execute
method does not ever get executed.
At this stage the whole thing consists of two units.
processthread.pas implements the TprocessThread class which is just a simple
TThread descendent which runs a command.
TRunInThread is a TComponent descendent which encapsulates TProcessThread with
a component, setting up an instantiation of it, executing it, storing and
updating the output etc.
In theory it should all work so I really do not know why it's throwing a
trace, perhaps somebody else can tell me what to change ?
I attach both the units for your viewing pleasure.
--
there's nothing as inspirational for a hacker as a cat obscuring a bug
by sitting in front of the monitor - Boudewijn Rempt
A.J. Venter
Chief Software Architect
OpenLab International
www.getopenlab.com
www.silentcoder.co.za
+27 82 726 5103
unit processthread;
{$mode objfpc}{$H+}
interface
uses
Classes, SysUtils,process;
Type
TProcessThread = class(tthread)
protected
procedure Execute; override;
public
fcommand : String;
Constructor Create(isSuspended : boolean);
end;
Var
MemStream :TMemoryStream;
Running:Boolean;
implementation
constructor TProcessThread.Create(isSuspended : boolean);
begin
MemStream := TMemoryStream.Create;
inherited Create(isSuspended);
end;
procedure TProcessThread.Execute;
const
READ_BYTES = 2048;
Var
Process :Tprocess;
n: LongInt;
BytesRead: LongInt;
begin
Process := TProcess.create(nil);;
Process.CommandLine := Fcommand;
Running := True;
{Actually run the thing and catch the output}
BytesRead := 0;
Process.Options := [poUsePipes,poNoConsole];
Process.Execute;
while Process.Running do
begin
// make sure we have room
MemStream.SetSize(BytesRead + READ_BYTES);
// try reading it
n := Process.Output.Read((MemStream.Memory + BytesRead)^, READ_BYTES);
if n 0
then begin
Inc(BytesRead, n);
end
else begin
// no data, wait 100 ms
Sleep(100);
end;
end;
// read last part
repeat
// make sure we have room
MemStream.SetSize(BytesRead + READ_BYTES);
// try reading it
n := Process.Output.Read((MemStream.Memory + BytesRead)^, READ_BYTES);
if n 0
then begin
Inc(BytesRead, n);
end;
until n = 0;
MemStream.SetSize(BytesRead);
Process.Free;
Running := False;
end;
end.
unit RunInThread;
{$mode objfpc}{$H+}
interface
uses
LResources,Classes, SysUtils,ProcessThread;
Type
TNewOutPutEvent = procedure(Sender: TObject;NewOutPut:TStringList) of object;
TRunInThread = Class(TComponent)
private
FOutPut : TStringList;
FNewOutPut :TStringlist;
FProcessThread : TProcessThread;
FCommand : String;
FRunning : Boolean;
FOnNewOuput : TNewOutPutEvent;
FOnDone : TNotifyEvent;
FLastSize : Integer;
FLastMSize : Integer;
procedure GetOutPut;
destructor destroy; override;
public
constructor Create(AOwner: TComponent);
procedure run;
published
property Command : String read FCommand write Fcommand;
property OutPut: TStringLIst read FOutput;
property Running: Boolean read FRunning;
property OnNewOutput : TNewOutPutEvent read FOnNewOuput write FOnNewOuput;
property OnDone : TNotifyEvent read FOnDone Write FOnDone;
end;
procedure Register;
implementation
procedure Register;
begin
RegisterComponents('OLPack',[TRunInThread]);
end;
Constructor TRunInThread.Create(AOwner: TComponent);
Begin
FLastSize := 0;
FLastMSize := 0;
FOutPut := TStringList.Create;
FNewOutPut := TStringList.Create;
FProcessThread := TProcessThread.Create(false);
FProcessThread.FCommand := FCommand;
if Assigned(FProcessThread.FatalException) then
raise FProcessThread.FatalException;
inherited create(AOwner);
end;
Destructor TRunInThread.Destroy;
Begin
FProcessThread.Suspend;
FOutPut.Free;
FNewOutPut.Free;
FProcessThread.Terminate;
inherited destroy;
end;
Procedure TRunInThread.GetOutPut;
Begin
FOutPut.LoadFromStream(MemStream);
end;
procedure TRunInThread.Run;
Var I : Integer;
Begin
FProcessThread.Resume;
FRunning := True;
While Running do
Begin
If MemStream.Size FLastMSize then
begin
FLastMSize := MemStream.Size;
GetOutPut;
If
Re: [lazarus] TRunInThread - Progress, but new problem
On Sunday 05 March 2006 08:43, A.J. Venter wrote: Well since my last mail on this a few weeks ago, I have made quite a bit of progress in my attempts to create a component that will allow me to call a tprocess in a thread while retrieving the output live. The structure I have no compiles, and can be placed on a form - it is all but right. The only trouble is - when I call the run method, it throws an access violation, the stack trace places it at line 76 (the line where I call FProcessThread.resume), I did a check and the first line of the execute method does not ever get executed. Mmmm, I think I know now why this is happening, though now I have even LESS of an idea how to prevent it. I instantiated TProcessThread directly in a test program (obviously this left me with no events or such but I was just trying to see if it worked) when I call TProcessThread.Create the program dies with error 232 complaining that there isn't a thread driver. I checked the lpr, it calls cthreads. I checked lazarus.pp - that too calls cthreads, the only thing I can imagine is that I must somehow call cthreads in olpack, putting it in olpack.lpk doesn't work - lazarus recreates the file on compile/save apparently, putting it in runinthread.pas doesn't work (probably because it's not early enough in the uses clause) - leaving me unable to determine where I SHOULD add it - or is this the IDE telling me that what I am trying to do is simply not possible ? TIA A.J. -- there's nothing as inspirational for a hacker as a cat obscuring a bug by sitting in front of the monitor - Boudewijn Rempt A.J. Venter Chief Software Architect OpenLab International www.getopenlab.com www.silentcoder.co.za +27 82 726 5103 _ To unsubscribe: mail [EMAIL PROTECTED] with unsubscribe as the Subject archives at http://www.lazarus.freepascal.org/mailarchives
