For finding leaks in HL1, is pointfile 100% accurate?

For finding leaks in HL1, is pointfile 100% accurate?

Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Orpheus on Fri Nov 24th 2006 at 8:40pm
Orpheus
13860 posts
Posted 2006-11-24 8:40pm
Orpheus
member
13860 posts 2024 snarkmarks Registered: Aug 26th 2001 Occupation: Long Haul Trucking Location: Long Oklahoma - USA
OK, this is bugging the s**t out of me because I am sure that I am right but, am willing to hear (not necessarily listen to and heed) others stories on the subject.

I believe that using pointfile correctly, and with enough particles added that it will always find your leak. Assuming of course that you have only one and not several. (It may overlook a second or third leak for which you may assume its not 100% accurate erroneously)

Tell me your stories if you found a situation where pointfile failed you in finding your leak.

Disclaimer: If you feel that pointfile isn't 100% accurate, can you please give specific details of what you did, or didn't do to get this result. Please don't just say, it failed.

The best things in life, aren't things.
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Crono on Fri Nov 24th 2006 at 8:46pm
Crono
6628 posts
Posted 2006-11-24 8:46pm
Crono
super admin
6628 posts 700 snarkmarks Registered: Dec 19th 2003 Location: Oregon, USA
It doesn't show all of them, but it will show "new" ones if you fix a leak and recompile and re-load the point file.

Of course it's accurate. The compiler knows there's a leak, so it makes of log of where it is. Hammer's implementation seems to show the traversal to the leak as well as give the nearest entity.

If it says there's a leak, there's a leak. Usually when people get frustrated with these they say it doesn't work, but there's some rules they're not aware of (like entities in the void)

It's 100% accurate because it's the same data that tells the compiler there's a problem in the first place. (Or is suppose to be)
Blame it on Microsoft, God does.
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Campaignjunkie on Fri Nov 24th 2006 at 8:49pm
Campaignjunkie
1309 posts
Posted 2006-11-24 8:49pm
1309 posts 329 snarkmarks Registered: Feb 12th 2002 Occupation: Student Location: West Coast, USA
Load it in Hammer - don't bother using the old HL1 pointfile method.

Anyway, it's always been 100% correct for me. Sometimes my puny human eyes wouldn't see the leak, but sure enough, it's always there.
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Stadric on Fri Nov 24th 2006 at 8:51pm
Stadric
848 posts
Posted 2006-11-24 8:51pm
Stadric
member
848 posts 585 snarkmarks Registered: Jun 3rd 2005 Occupation: Slacker Location: Here
I've had one problem with it that never got resolved. My 3D skybox was supposedly leaking, after I surrounded it with brush walls, and checked if the origin was in the void. Neither helped. What eventually did help was placing a large brush where the pointfile was pointing.

I'm sure there's an explanation, and I'd like to hear it.
Also change the texture of the dock. Docks are rarely tile. -Facepunch
As I Lay Dying
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Orpheus on Fri Nov 24th 2006 at 8:51pm
Orpheus
13860 posts
Posted 2006-11-24 8:51pm
Orpheus
member
13860 posts 2024 snarkmarks Registered: Aug 26th 2001 Occupation: Long Haul Trucking Location: Long Oklahoma - USA
I am confused. We have 3 of us here who said "YES" yet we have 3 votes and only one "YES"

How is that possible?

Anyway, keep'em coming gents. I wanna settle this once and for all.

The best things in life, aren't things.
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by reaper47 on Fri Nov 24th 2006 at 8:53pm
reaper47
2827 posts
Posted 2006-11-24 8:53pm
reaper47
member
2827 posts 1921 snarkmarks Registered: Feb 16th 2005 Location: Austria
It very rarely goes right through brushes (non-entity, of course). I don't know if that's an error of pointfile or BSP in general (mostly changing the brush the pointfile goes through fixes the leak, so it's hard to say).

I'd say BSP's leak-finding routine or brush-splitting isn't 100% accurate. But it's 99% accurate.
Why snark works.
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by ReNo on Fri Nov 24th 2006 at 8:57pm
ReNo
5457 posts
Posted 2006-11-24 8:57pm
ReNo
member
5457 posts 1991 snarkmarks Registered: Aug 22nd 2001 Occupation: Level Designer Location: Scotland
In HL1 I don't recall any problems with the pointfile, but the same can't be said for Source. Several times now one of my work mates and I have had pointfiles going straight through solid brushes completely unrelated to changes we have recently made, and had to rework geometry (not plugging leaks, just making changes that would normally be irrelevant like changing the brush layout for the same eventual result) in order to fix them. Quite often, you will make a change, recompile, and a new pointfile will be generated that goes straight through another, entirely seperate, perfectly legal brush. I've yet to find out the cause of these anomalies, but they are frustrating, and often the fix is something so ridiculously unrelated to leaks and the brushwork the pointfile was going through that you can't help but feel its simply buggy compile tools.
[img]http://card.mygamercard.net/sig/Default/reno84.png[/img]
Designer @ Haiku Interactive | ReNo-vation.net
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Mr.INSANE on Fri Nov 24th 2006 at 9:38pm
Mr.INSANE
156 posts
Posted 2006-11-24 9:38pm
156 posts 86 snarkmarks Registered: Jan 29th 2006 Occupation: Student Location: California,USA
I Use Quark and it loads pointfiles but most of the time it looks like the entire map is leaking with the line.

I use the block method to find leaks it has yet to fail me :smile:

For me pointfiles are useless
Why Do we all have to wear these ridiculous ties
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Pvt.Scythe on Fri Nov 24th 2006 at 9:53pm
Pvt.Scythe
730 posts
Posted 2006-11-24 9:53pm
730 posts 113 snarkmarks Registered: Sep 19th 2004 Occupation: student Location: Finland
<span style="color: white;">I've never used point files as far as I can remember, since the big block method is less of an hassle and easier to interpret(even when there are multiple leaks). Block method never fails and in some cases you can even leave the block where the leak was(for example the time I forgot to add roof to a room in cs_block).</span>
''Everyone wades in s**t until they're competent enough to walk on it. Jesus style.''
Dystopia - Empires
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by G.Ballblue on Fri Nov 24th 2006 at 10:59pm
G.Ballblue
1511 posts
Posted 2006-11-24 10:59pm
1511 posts 211 snarkmarks Registered: May 16th 2004 Occupation: Student Location: A secret Nuclear Bunker on Mars
I never really bother with pointfiles. Not only do they drain a hefty amount of resources on my PC, but in the time it takes for me to scan that huge-hunk-of-giberish line, I can generally find and plug the leak with Lep's big block method.
Breaking the laws of mapping since 2003 and doing a damn fine job at it
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Finger on Fri Nov 24th 2006 at 11:18pm
Finger
672 posts
Posted 2006-11-24 11:18pm
Finger
member
672 posts 1460 snarkmarks Registered: Oct 13th 2001
I dunno how the big block method could be any faster than pointfile, since it obviously takes more iterations to find the leak. 9 times out of 10 pointfile leads me right to the leak. Every once in a while, I get something weird like Reno suggested due to some crazy construction error, which the big block method wouldn't have been any more useful for solving.
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Orpheus on Fri Nov 24th 2006 at 11:33pm
Orpheus
13860 posts
Posted 2006-11-24 11:33pm
Orpheus
member
13860 posts 2024 snarkmarks Registered: Aug 26th 2001 Occupation: Long Haul Trucking Location: Long Oklahoma - USA
So, are you guys saying the pre-steam was 100% but post-steam its random?

dangit. :cry:

The best things in life, aren't things.
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Crono on Sat Nov 25th 2006 at 12:00am
Crono
6628 posts
Posted 2006-11-25 12:00am
Crono
super admin
6628 posts 700 snarkmarks Registered: Dec 19th 2003 Location: Oregon, USA
pre and post steam doesn't matter, the compile tools only changed for the game, not the interface in which you play the game.

I've had less problems with the point files in hl2, to be honest. I have yet to encounter the "not really a problem, but it still shows up as one" leak. But I don't use it on a daily basis, nor is it part of my job.
Blame it on Microsoft, God does.
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Captain P on Sat Nov 25th 2006 at 12:12am
Captain P
1370 posts
Posted 2006-11-25 12:12am
1370 posts 1995 snarkmarks Registered: Nov 6th 2003 Occupation: Game-programmer Location: Netherlands
I hardly ever used pointfiles. I learned how to prevent leaks mostly, but besides that, I can usually find leaks by just looking through a map in Hammer (after deleting all entities and clip/hint brushes). And in those peculiar cases, the big block method does the trick. As long as you don't do a full recompile it's not too slow either.

But perhaps my ignorance of pointfiles is because I never loaded them into Hammer. Looking after them in Half-Life isn't all that effective imho. :smile:
Create-ivity - a game development blog
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Gorbachev on Sat Nov 25th 2006 at 1:12am
Gorbachev
1569 posts
Posted 2006-11-25 1:12am
1569 posts 264 snarkmarks Registered: Dec 1st 2002 Location: Vancouver, BC, Canada
Pointfiling in Hammer takes me all of like 5 seconds to find the problem. It's always been a snap, and a godsend so you don't have to do big block or in-game methods.

I have had once similar to what Reno's talking about with the original HL engine where it had a leak through a regular joe non-entity block that I hadn't touched in forever. I tried reshaping, resizing, etc. to no avail, just deleted and remade the exact same block and it magically was all better.
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by fishy on Sat Nov 25th 2006 at 1:38am
fishy
2623 posts
Posted 2006-11-25 1:38am
fishy
member
2623 posts 1476 snarkmarks Registered: Sep 7th 2003 Location: glasgow
loading the pointfile in hammer to find normal leaks, as well as areaportal problems, is so much better than the old method of running the map, though i've had odd errors with both old and new.
i eat paint
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Dr Brasso on Sat Nov 25th 2006 at 9:19pm
Dr Brasso
1878 posts
Posted 2006-11-25 9:19pm
1878 posts 198 snarkmarks Registered: Aug 30th 2003 Occupation: cad drafter Location: Omaha,NE
did i miss this, or does no one use the cordon tools?.. :eek:

and totally dependant on the situation, the point file is acurate enough....alternative methods are sometimes necessary jon....handy to have

Doc B... :dodgy:
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Orpheus on Sat Nov 25th 2006 at 9:29pm
Orpheus
13860 posts
Posted 2006-11-25 9:29pm
Orpheus
member
13860 posts 2024 snarkmarks Registered: Aug 26th 2001 Occupation: Long Haul Trucking Location: Long Oklahoma - USA
Did they even have the cordon tool in HL1?

Anywho, this was a thread primarily to establish pointfiles accuracy. There are plenty of other methods, up to and including not creating holes, for finding them.

I was just curious to see how, or if anyone had any instances where the method wasn't 100% accurate.

So far, the jury seems out to lunch, since I cannot count those people who never use it, or used it incorrectly. Other than this new hole in a solid wall thing anyway.

The best things in life, aren't things.
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Captain P on Sat Nov 25th 2006 at 11:00pm
Captain P
1370 posts
Posted 2006-11-25 11:00pm
1370 posts 1995 snarkmarks Registered: Nov 6th 2003 Occupation: Game-programmer Location: Netherlands
I've had a few troublesome issues with the cordon tool once, for HL (loosing the rest of my map somehow), so I'm not too happy with trying it again. Well, as long as you keep your .rmf files safe it shouldn't give any problems, as it just controls what goes into the .map file, but still.

Your grinny emoticon is funny btw, Brasso. :smile:
Create-ivity - a game development blog
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Le Chief on Sun Nov 26th 2006 at 2:52am
Le Chief
2605 posts
Posted 2006-11-26 2:52am
Le Chief
member
2605 posts 937 snarkmarks Registered: Jul 28th 2006 Location: Sydney, Australia
u know what.I have always wanted to know how to do the whole point file thing. How does it work anyway.

DANG
Aaron's Stuff
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by omegaslayer on Sun Nov 26th 2006 at 7:18am
omegaslayer
2481 posts
Posted 2006-11-26 7:18am
2481 posts 595 snarkmarks Registered: Jan 16th 2004 Occupation: Sr. DevOPS Engineer Location: Seattle, WA
This is strictly for HL1 aaron (This topic). If you have a leak in your
map you load it in the game and in the console type "pointfile" and it
should lead you to the general direction of the leak. If you want to
know how to do it in HL2 you compile your map (find a leak) and go
map-->load point file. Then follow the red line to the hole in the
map.

To replay to orph. Its not 100% accurate, but it does give you a
general idea of where the leak is (if your comp can run a leaked map).

In HL2 its a god-send tool.
Posting And You
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by mazemaster on Tue Nov 28th 2006 at 12:14am
mazemaster
890 posts
Posted 2006-11-28 12:14am
890 posts 438 snarkmarks Registered: Feb 12th 2002
Hammre can only display a certain maximum number of points. If the pointfile has more than this, then random points are removed. Thus if you have a small leak near the edge of your wall, and the points near the edge are randomly removed, then the pointfile line in Hammre will go straight through the wall into the void instead of going through the gap.

In-game you can always avoid this problem by specifing more points.
http://maze5.net
Re: For finding leaks in HL1, is pointfile 100% accurate? Posted by Fjorn on Wed Nov 29th 2006 at 10:00pm
Fjorn
250 posts
Posted 2006-11-29 10:00pm
Fjorn
member
250 posts 25 snarkmarks Registered: Jun 5th 2004 Occupation: Student/Amateur Writer Location: California - USA
I've had the pointfile go in squares in the middle of an area that was part of the map

rather odd
Signature? What signature!?