Author Topic: "Is Not a Valid Win32 Application" when installing - WHAT'S HAPPENING? HELP! (Read 47,079 times)

mouser · « **on:** January 04, 2008, 08:26 PM »

I'm getting a continual trickle of users who are getting this error when trying to download and run the installers to my programs.

The proximal cause of the problem is very clear: It happens when the setup file does not download completely.

It also only seems to happen with Internet Explorer (maybe only version 7? not sure).

The real question is, why on earth is Internet Explorer messing up and only downloading part of the file and then thinking it has downloaded it all?

And how can I fix this?
(yes i know i know, i already recommend FireFox to people who experience the problem but it's not the perfect solution).

jgpaiva · « **Reply #1 on:** January 04, 2008, 08:32 PM »

Opera!

Just kidding.

Are you sure that iexplorer says it has finished downloading? Couldn't it be that the server is stopping the downloads before they are finished? I'm not sure what would happen in that case, but maybe iexplore would say it was done.

tranglos · « **Reply #2 on:** January 04, 2008, 08:53 PM »

This won't help much, but I've just downloaded ProcessTamer and Screenshot Capror successfully using IE7 on XP SP2. Both installers ran fine. I thought also maybe the server was reporting a wrong file size, and checked that too, but no.

Did you ask if the people having this problem are running an AV scanner or any similar software? A common firewall maybe?

Also, you might try checking your server logs, though I'm not sure if Apache logs any information that would identify aborted downloads or disconnects. Perhaps it's worth a try.

f0dder · « **Reply #3 on:** January 04, 2008, 11:58 PM »

Hmmm... sounds weird. Normally with installers, there's a .exe stub and the archive is simply appended to that - with such an approach, unless the download stops/is corrupted very early, you shouldn't get bitching about "not a valid win32 application". If your installer embeds the archive as a Win32 PE resource, then it's a different issue, but I don't think it does since WinRAR can extract from the installers...

I doubt it's a MIME filetype association thing, but that's perhaps something to investigate as well; if the browser thinks it's being served a text file, it might do LF->CRLF translation, and that would certainly mess up things. Perhaps you could ask one of the people experiencing the error to email you the setup archive, and you could compare the first few kilobyte of it with an uncorrupted install file?

tranglos · « **Reply #4 on:** January 05, 2008, 01:14 AM »

Quote from: f0dder on January 04, 2008, 11:58 PM

I doubt it's a MIME filetype association thing, but that's perhaps something to investigate as well; if the browser thinks it's being served a text file, it might do LF->CRLF translation, and that would certainly mess up things. Perhaps you could ask one of the people experiencing the error to email you the setup archive, and you could compare the first few kilobyte of it with an uncorrupted install file?

Could be, but DC http server indicates content type correctly. This is the header sent for http://www.donationcoder.com/Software/Mouser/screenshotcaptor/downloads/ScreenshotCaptorSetup.exe:

hdr>Date: Sat, 05 Jan 2008 01:07:12 GMT hdr>Server: Apache hdr>Last-Modified: Mon, 08 Oct 2007 02:38:33 GMT hdr>ETag: "5d0139-886000-47099829" hdr>Accept-Ranges: bytes hdr>Content-Length: 8937472 hdr>Connection: close hdr>Content-Type: application/octet-stream

Mouser, how many reports of this did you get? Are those IE7s running on Vista?

cmpm · « **Reply #5 on:** January 05, 2008, 02:15 AM »

Just guessing here, but I'd say they may be hitting run instead of save. So it could have something to do with it being in the tempoary folder and installed from there. That's the only difference I can think of at this point.

cmpm · « **Reply #6 on:** January 05, 2008, 04:12 AM »

The main solution after investigating via google is to clear the cache.
Has this been tried?

Here's a good cleaner for more then the cache even.

http://www.atribune.org/content/view/25/2/

mouser · « **Reply #7 on:** January 05, 2008, 05:36 AM »

interesting ideas cmpm..

as you say, people seem to suggest that this happens and you have to clear the IE cache to get it to be willing to redownload the file..

but this doesn't really tell you WHY IE gets into this confused state of THINKING it's downloaded the whole file, when normally it does not. in fact the cache issue just really makes the problem 10x worse since it makes it impossible for the user to just redownload after they get this error the first time! (thank you bill gates!)

i guess what i'm trying to get at is that normally in IE when you try to download a file, it KNOWS when it is interrupted midstream. it only moves the file from temp dir to real download directory after it finishes downloading the file.

but in these rare cases, clearly IE *thinks* the download has completed, and gives it to the user to run. so this is still a mystery as to what event or condition is causing IE to get confused about this.

cmpm · « **Reply #8 on:** January 05, 2008, 05:55 AM »

yeah lol...I think ie7 got dumber in some ways then ie6.
built to browse the web basically and not really good at much else

could be a combination of a few normal things

i wonder if they are using that ie7pro which has gone to version 2 now
and a top rated download at ms

Deozaan · « **Reply #9 on:** January 05, 2008, 06:00 AM »

I'm not sure if this is the same thing, but occasionally a large file will be downloading and it will only get a few MB into it and it will stop, saying it was successful. This has happened to me in Firefox too.

It tends to do it to me when I'm downloading many files at once and/or my connection is going really slow.

mouser · « **Reply #10 on:** January 05, 2008, 06:02 AM »

cmpm, i will ask them next time if they are using any downloaders.. that might be a possible explanation also.
i guess i was hoping it was some misconfiguration on our server which could be corrected.

nosh · « **Reply #11 on:** January 05, 2008, 09:17 AM »

Check this out.
This guy's Apache log shows the full file going through, the fault lies completely with IE.

The consolation prize is that you can use these opportunities to move these people towards better browsers.

mouser · « **Reply #12 on:** January 05, 2008, 09:24 AM »

nosh, that's a great find -- that's definitely the problem and it's nice to read someone's attempt at diagnosing it.
bad news is.. he never found a solution.

mouser · « **Reply #13 on:** January 05, 2008, 09:26 AM »

more info: http://www.fourmilab.ch/documents/corrupted_downloads/

i wonder if one possible solution is to add some extra ?cruft to the download url to prevent ie from cacheing it?

cmpm · « **Reply #14 on:** January 05, 2008, 09:46 AM »

you would either have to tell them to disable caching or it be done auto matically with the download when hitting setup

which is out of my realm of know how

here's some ms stuff about disabling it

http://search.microsoft.com/results.aspx?mkt=en-US&setlang=en-US&q=Disable+caching

nosh · « **Reply #15 on:** January 05, 2008, 10:07 AM »

My understanding is clearing the cache will only give the user a shot at re-downloading it. It won't prevent IE fudging up the first time.

Off the top of my head:
FTP hosting
Hosting at a site that uses Activex or whatever to let IE users resume broken downloads - the ones with the download progress showing right within the web page.
Creating a tiny stub file of your own, one that connects to your server and makes sure the whole stream goes through, yes they'll have to download the stub first... *sigh!*

f0dder · « **Reply #16 on:** January 05, 2008, 10:25 AM »

Ugh, please don't add ?cruft just because one browser is horribly broken >_<

app103 · « **Reply #17 on:** January 05, 2008, 10:32 AM »

This is a known IE issue...and it doesn't just affect IE 7...it was a problem with IE 6 too.

It also affects most IE based browsers such as Maxthon and AOL's various browsers.

On your end of things, there isn't anything you can do to fix IE for users.

On their end, however, there are a couple of solutions they can use that will help them on most sites and not just DC:

1. Use a different browser such as Firefox, Opera, K-Meleon, etc (something not IE based)
2. Use a download manager of some sort. (this is what I opted for)

Sites like AOL opted for another option, being that they are quite aware of the problem and it affects users of their IE based browser and has caused problems with their customers downloading new versions of their software.

They issue a small application that is tiny enough to download fully without problems. This tiny application when run, will download the real software (escaping the IE problem), always the latest version, and install it for the user. Their tiny downloader application is also pausable and can resume a broken download.

I am not sure of the file size threshold for escaping the broken download risk in IE. That would be something you would have to find out and then keep your downloader app's file size under that amount. (if you were to go with this option)

f0dder · « **Reply #18 on:** January 05, 2008, 10:36 AM »

Downloader might be a good idea, as long as you also keep the regular files downloadable; I hate when I'm forced to use a custom, and I'm sure I'm not the only one.

Lashiec · « **Reply #19 on:** January 05, 2008, 08:48 PM »

I'd suggest assembling a giant mob and march over Redmond ("Fix that bug already, you lazy slowassed engineers"

)

Stoic Joker · « **Reply #20 on:** January 06, 2008, 05:21 AM »

Quote from: mouser on January 05, 2008, 06:02 AM

cmpm, i will ask them next time if they are using any downloaders.. that might be a possible explanation also.
i guess i was hoping it was some misconfiguration on our server which could be corrected.

Server Administration 101 - Never make changes to the server based on user input. Check the logs, if the server says it's OK ... then 99% of the time it is. Feel free to adjust users as necessary...

Granted that's for LAN/WAN network administration, but it transcends rather nicely to public access services.

One thing that is not being addressed here is how much control do you have over "your server". My web server is sitting in a rack about 10' from me ... so I have complete control. If yours is at a hosting company (dedicated or not...) you're at their mercy as to how to handle traffic spikes. Transfer stability can be impacted depending on how the hosting company decides to "throttle back" a site during a spike.

Blaming IE for the DL issues is convenient (easy... as IE bashing appears to be a hobby round these parts) but it doesn't resolve anything because it doesn't really identify anything specific. But then again witch hunts seldom do.

I spend a goodly part of my day explaining to users why their computers don't really hate them, and removing tons of 3rd party (tripe) browser add-ons, tool-bars, "Helper Objects", and all manner of other crap that they thought they just had to have.

I've always used IE and never had a problem... however I have seen many things break it and cause issues. Just for fun, while exploring the client side aspect of the issue, ask the broken download crowd how many of them are using the Norton Internet Security suite. Symantec got in a fight with MS awhile back over the smooth-wall (no kernel hooks) kernel ... and their over compensation for it has turned all their AV products into total poo.

f0dder · « **Reply #21 on:** January 06, 2008, 12:26 PM »

FWIW I've had the "weird download corruption" problem every now and then with IE (though never from this site), and I didn't use any browser extensions except for flash and the windows-update stuff, I didn't have antivirus software running, etc. Don't think I've ever seen the problem with other browsers, they've notified me if something went wrong while downloading.

<offtopic>Oh, and shame on MS for not providing an alternative to kernel hooking after they introduced PatchGuard</offtopic>

Stoic Joker · « **Reply #22 on:** January 06, 2008, 06:39 PM »

Quote from: f0dder on January 06, 2008, 12:26 PM

FWIW I've had the "weird download corruption" problem every now and then with IE (though never from this site), and I didn't use any browser extensions except for flash and the windows-update stuff, I didn't have antivirus software running, etc. Don't think I've ever seen the problem with other browsers, they've notified me if something went wrong while downloading.

My angle on this is that it's a Chicken-or-the-Egg" story. IE has some fault for lacking user feedback depending on how the sending server resets the connection. The server has some fault for resetting the connection. Who is more guilty-er *shrug* is a matter of opinion. ...But that (cause/effect scenario) IMO is the answer to the WTF is happening question.

Quote

<offtopic>Oh, and shame on MS for not providing an alternative to kernel hooking after they introduced PatchGuard</offtopic>

Offtopic=true so I PM'ed response.

app103 · « **Reply #23 on:** January 06, 2008, 10:20 PM »

I have experienced this IE related problem too, beginning shortly after upgrading to IE 6 on my other machine.

No browser add-ons there, not even flash. Nothing Norton on that machine either.

Same IE problems with this machine, and my daughter's laptop. And I know plenty of other people that have experienced it, some much worse than others.

Upgrading to IE 7 doesn't fix it. If you have a way of downgrading to IE 5.5, it will fix it, but that isn't really a wise thing to do for security reasons.

This is a known IE issue.

lanux128 · « **Reply #24 on:** January 07, 2008, 12:43 AM »

imo, it's easier to suggest to users to use a download manager due to the size of the program (e.g. ScreenshotCaptorSetup.exe = 8.52MB). that way we don't have pull our hairs over a Microsoft problem..