Lag in detect change in included files with $INCLUDE
lazarus / 21-General

----------------------------------------------------------------------
alantelles · 2021-01-24 19:32

Hello, y'all.

Many times, when I make changes to included sources and compile
program the changes are not "taken account". The program is compiled
without the changes I've done. Only after some time or some random
"close-open" Lazarus the program is rightfully compiled. There is a
logic in this behaviour or something I should have done that I didn't?

Edit: FPC 3.2.0, Lazarus 2.0.10, however it happened also in FPC 3.0.4
with Lazarus 2.0.6

----------------------------------------------------------------------
jamie · 2021-01-24 20:06

did you try "Clean Up and Build all " ?

----------------------------------------------------------------------
MarkMLl · 2021-01-24 20:41

OP: please tell us EXACTLY what version of compiler etc. you're using.
That would be important under normal circumstances, but in this case
you might be looking at a recently-introduced compiler bug.

Using FPC 3.2.0 and Lazarus 2.1.0 on Linux x86_64 I think I've been
seeing something comparable.

Using this code:

Code: Pascal  [Select][+][-]window.onload = function(){var x1 =
document.getElementById("main_content_section"); if (x1) { var x =
document.getElementsByClassName("geshi");for (var i = 0; i program
PythonDemo; {$mode objfpc}{$H+} { define DYNAMIC }{$define PYTHON2
} ... uses  {$IFDEF UNIX}  cthreads,  {$ENDIF}  Classes,{$ifdef
DYNAMIC }  python_dynamic;                       (* "Python" is a
shared-library object  *){$note ===== Intending to link dynamically
===== }{$else          }  python;                               (*
"Python" is a unit file              *){$note ===== Intending to link
statically ===== }{$endif DYNAMIC }... 
if I change the define DYNAMIC state I see the main unit (excerpted
above) recompile, but there appears to be no change of what file is
actually linked... Lazarus's "Build all" doesn't fix it, it's
necessary to do a "Clean directory". I've not yet tried to work
through older versions.

MarkMLl

----------------------------------------------------------------------
alantelles · 2021-01-25 11:26

Quote from: MarkMLl on January 24, 2021, 09:41:56 pmOP: please tell us
EXACTLY what version of compiler etc. you're using. That would be
important under normal circumstances, but in this case you might be
looking at a recently-introduced compiler bug.

Using FPC 3.2.0 and Lazarus 2.1.0 on Linux x86_64 I think I've been
seeing something comparable.

Using this code:

Code: Pascal  [Select][+][-]window.onload = function(){var x1 =
document.getElementById("main_content_section"); if (x1) { var x =
document.getElementsByClassName("geshi");for (var i = 0; i program
PythonDemo; {$mode objfpc}{$H+} { define DYNAMIC }{$define PYTHON2
} ... uses  {$IFDEF UNIX}  cthreads,  {$ENDIF}  Classes,{$ifdef
DYNAMIC }  python_dynamic;                       (* "Python" is a
shared-library object  *){$note ===== Intending to link dynamically
===== }{$else          }  python;                               (*
"Python" is a unit file              *){$note ===== Intending to link
statically ===== }{$endif DYNAMIC }... 
if I change the define DYNAMIC state I see the main unit (excerpted
above) recompile, but there appears to be no change of what file is
actually linked... Lazarus's "Build all" doesn't fix it, it's
necessary to do a "Clean directory". I've not yet tried to work
through older versions.

MarkMLl

Edited to tell compiler version

----------------------------------------------------------------------
MarkMLl · 2021-01-25 11:33

Could you post a simple test project which demonstrates the problem?

MarkMLl

----------------------------------------------------------------------
alantelles · 2021-01-25 18:51

I think it would not be able to reproduce because i think it's a
"progressive" problem. As you use more included files the issue
appears. It didn't happen until I have some more included files. There
would be a lot of steps to reproduce in my project however, if you're
realy interested I can leave it here

----------------------------------------------------------------------
MarkMLl · 2021-01-25 19:23

I'll try to take a bit of a look at it myself in the morning, and will
report back.

MarkMLl

----------------------------------------------------------------------
MarkMLl · 2021-01-31 15:04

My apologies for the delay looking at this. I've just been tinkering
with simple include and unit files in the same directory using the
attached test files without finding anything untoward. @alantelles
Please could you see if you can duplicate your problem using some
variant of these, and if not I think we could usefully start looking
at the unitpath directive etc.

MarkMLl