[lazarus] Just another bug

[Mattias Gaertner]

On Wed, 06 Nov 2002 17:31:18 +0300
"Vasily I. Volchenko" <Vasily.Volchenko at mstu.edu.ru> wrote:

> 
> 
> >
> > The gdb backtrace shows that the RTTI is destroyed. That's bad. It can
> > mean that the heap is destroyed.
> 
> Bug was fixed at changing rtl,

ok


> but last lazarus versions have rejected the fpc path.

Rejected? What was the path and what was the error message?


> So, it was impossible to make an event. I tried building them with
> different rtls, but nothing helped. It finded lazarus dir, but not an FPC
> dir, written corretly (and even incorrectly :) ) in Env. options. The
> first was SysUtils unit. When I placed its path in line with CodeTools
> opts (in Other sources path), it forwarded to unit classes.

Perhaps it helps, if I explain what is done with the given paths:

The codetools has a built in heuristic:
First it asks the compiler to get the markros and unit paths. These unit
paths normally point to directories like /usr/lib/fpc/1.0.7/units/linux/...
.
Then it searches for all .ppu files in the unit paths.
Then it searches all .pp/.pas files in the fpc source path and creates an
index.

When the IDE starts up and whenever you change the fpc compiler or source
path in the environment options the IDE checks if the compiler executable
has changed. If it has changed the unit index is rebuild.

Because the fpc sources are full of IFDEFs the codetools can't parse them
without the makros. So, the compiler filename _must_ be valid.


> Should I place all needed fpc .pp - .inc files to another dir? 

No. It's probably just a typo.


> Or it is just a bug?

Maybe. But I need more information.
The exact paths in the environment options.


> Besides, very last version (yesterday) messaged only Abort without unit. 

Just 'Abort'? 


> Two week's version said the unit name.


Mattias