[Lazarus] suggestion for lazbuild

[Mehmet Erol Sanliturk]

On Wed, Mar 18, 2015 at 6:06 PM, Mattias Gaertner <nc-gaertnma at netcologne.de
> wrote:

> On Wed, 18 Mar 2015 17:22:25 -0700
> Mehmet Erol Sanliturk <m.e.sanliturk at gmail.com> wrote:
>
> >[...]
> > > > but it is not taken into consideration the record type changes
> mentioned
> > > > above .
> > >
> > > Are the altered units in the project or in a package?
> > >
> > >
> >
> > Altered units are in a project ( I never use any "package" ) , not loaded
> > into Lazarus GUI , but their directories are defined .
>
> Then I have no clue how to reproduce this. You have to create an
> example. Is there already a bug report?
>
>
>







My set up is like the following :


(1)

/dir_one/dir_two/type_source.pas

contains :

type numbers = record

     k : Integer ;

     end ;



(2)

/dir_three/dir_four/type_unit.pas

contains :

unit type_unit ;


interface


{$I /dir_one/dir_two/type_source.pas }

implementation

end .



(3)

/dir_three/dir_four/source_unit.pas

contains :

unit source_unit ;


uses type_unit ;



interface



implementation


{$I /dir_one/dir_five/compute_source.pas }



end .


(4)


/dir_one/dir_five/compute_source.pas

contains :


procedure compute ;

var V : type_unit . Numbers ;


begin


  v . k := 0 ;

end ;




(5)


There is a program in



/dir_six/dir_seven/computer.lpi


through its local units ,
one of the units

uses source_unit ;


and a procedure uses


   ...
   compute ;
   ...


and computer.lpi is compiled and executed succeefully .

All of the compiled units are stored into


/dir_eight/dir_nine/





(6)

Assume that a new element is included into record Numbers as

  m : Integer ;



(7)


A new statement included into

procedure compute as

  v . m := 0 ;



(8)


Computer.lpi is compiled in "Make" mode :


error is like

  v . m : Undefined  element



because type_unit is NOT compiled , but existing .ppu and .o is used .



(9)

During compilation with Lazarus , only

Computer.lpr

is loaded , and NO any other parts .



As a summary :



/dir_one/dir_two/type_source.pas
/dir_one/dir_five/compute_source.pas


/dir_three/dir_four/type_unit.pas
/dir_three/dir_four/source_unit.pas


/dir_six/dir_seven/computer.lpi


/dir_eight/dir_nine/ : Compiled units





At present , all of the directories are relative as ../  ...
but , when directories are absolute , the same error is displayed .


This situation is persisting at least since around 2007 .



( I do not remember which subject , during entering a bug report , the (
only report page ) is disappeared from my browser and it could not be
possible re_enter any more anything into that page . This was my last bug
report attempt )





> >[...]
> > > 17 minutes? Even the 3+ millions lines of FPC and Lazarus normally only
> > > takes a few minutes on a recent machine.
> >
> > Yes : Seventeen minutes  ( without debug related additions ) .
> >
> > Processor :  Intel i5-3470 , Quad-core . 3.2 GigaHerz
>
> How big is that project?
>
>
> > Alongside the ( NOT detecting record element additions / changes
> /deletions
> > ) , another very important problem for me is to
> >
> > ( Discontinuation of compilation on errors )
> >
> > although in an item "Display error messages up to 1 000 000 ( million )"
> is
> > selected .
> >
> >
> > This number is NOT taken into consideration , and it is stopping
> > compilation at the end of a unit or after a small number ( around 50 ) .
>
> For me FPC is so fast, that I never needed that feature.
> Please create a bug report.
>
> Mattias
>
> --
> _______________________________________________
> Lazarus mailing list
> Lazarus at lists.lazarus.freepascal.org
> http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus-ide.org/pipermail/lazarus/attachments/20150320/8e8684e4/attachment-0003.html>