Crash on Load and Fonts revisted?

[Beaverhunter]

Oki... well I played yesterday and it worked just fine with CVS but today it's not working. I assume that I compiled CVS and then just copied it over after I exited EL and went to bed and so on without testing the client, because now I'm crashing. After login I just crash right away.

 

I can probably get some debug info on this if I just could get the CVS to compile.

I'm currently getting this:

1>widgets.c

1>.\widgets.c(1482) : error C2275: 'text_field' : illegal use of this type as an expression

1>.\widgets.h(142) : see declaration of 'text_field'

1>.\widgets.c(1482) : error C2065: 'tf' : undeclared identifier

1>.\widgets.c(1485) : error C2223: left of '->buffer' must point to struct/union

1>.\widgets.c(1485) : error C2223: left of '->msg' must point to struct/union

1>.\widgets.c(1485) : error C2168: 'floor' : too few actual parameters for intrinsic function

1>.\widgets.c(1486) : error C2223: left of '->nr_visible_lines' must point to struct/union

1>.\widgets.c(1486) : error C2223: left of '->y_space' must point to struct/union

 

Maybe getting the current CVS to compile solves it, but well my CVS that seems to crash is from 2007-06-12 (about 11,5 hours ago).

 

Font issue:

Well when I had the CVS working yesterday, I went to check out the fonts since there has been much talk about them lately. When I switch to any other but Type 1, I get just a mess (like those screenshots seen before). The text is not readable.

 

I did download 1.4.0. client from the page as of yesterday and this did not make any difference BUT when using the official client I can see the fonts, with CVS not. (And yes I do have FONTS_FIX)

[Grum]

Re: widgets.c

Check if the CVS merge was successful, when in doubt, please post the entire offending function here (plus line numbers). I don't see anything suspicious around line 1482, so either the update went wrong, or your compiler is drunk.

[Beaverhunter]

Re: widgets.c

Check if the CVS merge was successful, when in doubt, please post the entire offending function here (plus line numbers). I don't see anything suspicious around line 1482, so either the update went wrong, or your compiler is drunk.

 

I would probably say that it's my compiler that allready has it's own definition of text_field and when using it like that and having an own struct called the same... I'm pretty sure my compiler complains. I'm not sure though, but I tried with newest CVS.

I use MSVS 2005 btw.

 

Update:

Oki, I rolled back to 12 june 2007 20:00 GMT+1 (8 pm).

And it compiles fine and runs just fine.

The one which crashed was compiled/changed/created 12 june 2007 21:59 GMT+1 (9:59 pm).

Now to see if the current CVS crashes or not... I need to get the above mentioned error in widget.c and widget.h fixed.

Font problem still remain.

Edited June 13, 2007 by Beaverhunter

[Grum]

I would probably say that it's my compiler that allready has it's own definition of text_field and when using it like that and having an own struct called the same...

I doubt it. Now could you check the actual files to see if the merge succeeded?

[Beaverhunter]

I would probably say that it's my compiler that allready has it's own definition of text_field and when using it like that and having an own struct called the same...

I doubt it. Now could you check the actual files to see if the merge succeeded?

 

How can it not have merged properly? The CVS uses hash checks to see if things are right, and I tried to remove the files and refetch them and still same errors with widget. Usually my compiler highlights things that are system defined with a pretty bright blue text (like a cyberlink on an average homepage). And text_field is probably a part of the .net, at least it wouldn't suprise me...

 

Ooh btw, I just searched references and I saw that the text_field is used in other files that doesn't give errors so hmm maybe I'm wrong about the define thing... but well the errors remain...

Edited June 13, 2007 by Beaverhunter

[Grum]

How can it not have merged properly?

By you making changes that conflict with the current tree. (not saying you did, just pointing out how conflicts could arise.

The CVS uses hash checks to see if things are right, and I tried to remove the files and refetch them and still same errors with widget.

Ok, so that's not the problem... oh, I think I found it. Thank MSVC for it clear error messages....not...

Please update CVS and see if it fixes your problem.

Usually my compiler highlights things that are system defined with a pretty bright blue text (like a cyberlink on an average homepage). And text_field is probably a part of the .net, at least it wouldn't suprise me...

It would surprise me, and it would surprise me even more if it caused this particular error.

[Beaverhunter]

Aah nice!

It seems to work now with updated CVS, now for testing for crashes or not...

 

I did however try to change the code to this:

 

if (w != NULL)
{
	text_field* tf = w->widget_info;
	float displayed_font_y_size = floor (DEFAULT_FONT_Y_LEN * tf->buffer[tf->msg].wrap_zoom);
	tf->nr_visible_lines = (int) ((w->len_y - 2*tf->y_space) / displayed_font_y_size);
}

 

And that compiled aswell, maybe it's not doing the same thing though but well, guess you fixed it.

 

Update:

Oki, the newest CVS compiled didn't crash on load.

___Font issue still present.___

Edited June 13, 2007 by Beaverhunter