[Lazarus] App locks at startup

[SteveG]

On 28/02/10 13:35, SteveG wrote:
> On 28/02/10 13:18, SteveG wrote:
>> On 28/02/10 13:08, Andrew Brunner wrote:
>>> I would never Process Application Messages from a timer thread.
>>>
>>> I suspect a threadlock going on. Does it work OK without the
>>> application.processmessages?
>>>
>>> On Sat, Feb 27, 2010 at 9:02 PM, SteveG<steveg at nevets.com.au> wrote:
>>>> I have just updated Laz (svn) and Fpc(2.4.0) and have hit a problem
>>>> with my
>>>> program. (worked ok with Laz 0.98 / fpc 2.?.?)
>>>>
>>>> I use a dyn loaded lib with my app, which seems to be relevant.
>>>>
>>>> If I call Application.ProcessMessages from within a TTimer.OnTimer
>>>> event, my
>>>> app now locks solid showing 'futex_wait_queue_me' in the process
>>>> listing.
>>>>
>>>> I havent yet been able to narrow down an exact cause, as a simple
>>>> app/lib
>>>> test works ok.
>>>>
>>>> Would anybody know what would cause the above state (futex), as a
>>>> clue to
>>>> the change in Laz/Fpc affecting my app ?
>>
>> It does seem to work without the App.proc call. The problem I have seems
>> to be caused by calling App.proc once my lib is loaded. This was the
>> closest I have been able to get in narrowing down to a small test case.
>> I call App.procMsgs from within my lib which is where the actual problem
>> seems to be.
>>
>> Is there any other way to create a 'pause' from within a
>> library/function call (not sleep, as it 'stops' the program).
>> My 'pause' function is works as :>
>> StTm := NOW;
>> while NOW < (StTm + TmOut) do begin
>> Sleep(1);
>> Application.ProcessMessages;
>> end;
>>
>> This worked ok before the Laz/fpc update, but seems to be broken now ?
>> This is just my 'guess' so far as where my problem lies ?
>>
> After more study (read as: more google) - it may be related to the
> kernel changes in Ubuntu 10.04 - seems to be many unrelated programs
> suffering similar problems - will continue this story as I find more :)
>

Just ran my app on Ubuntu 9.10 - works fine - so once again Laz/fpc has 
been falsely accused :)

Sorry for the noise