1. filesize error
- Posted by useless_ Nov 30, 2012
- 1918 views
- Last edited Dec 01, 2012
I'm lazy, here's the irc log of me talking about it...
[20:24] * any33068445 looks alarmed at Matt [20:25] <any33068445> the writefile opened as "ab" is not going above 2^32 exactly as reported by winxp (explorer and dosbox), but i used dos type command to append a 2.6G file to another 2.5G file, and it is reported to be 5.2G, as it should [20:26] <any33068445> it's the Eu file that's not being reported as the size it should be [20:58] <any33068445> ok, i started a eu app to make a 5.2G file, one line ata time .. 9 pm Friday 11/30/2012 .. .. 10 pm Friday 11/30/2012 .. [22:01] <any33068445> batman, we have a problem [22:02] <any33068445> this filesize has also stopped increasing at : 4.00 GB (4,294,967,296 bytes) [22:02] <any33068445> 2^32 = 4,294,967,296The app reported this on terminating:
[22:03] <any33068445> running [22:03] <any33068445> Before closing: 4294967296 [22:03] <any33068445> After closing: 4294967296 [22:03] <any33068445> Press Any Key to quit
Explorer reports filesize as 2^32 bytes
Dos dir reports it as 2^32 bytes.
Code used is at : http://openeuphoria.org/pastey/173.wc
I also used dos command "type" to append to a file, using the same 2.6G readfile and same writefile name, it correctly built a 5.2G (reported) file.
ADDED: I was using:
Euphoria Interpreter v4.0.4 Windows, Using Managed Memory Revision Date: 2012-04-02, Id: 11a49820121f
ADDED part 2: C:\euphoria40b4-3412\bin\eui.exe also truncated at 2^32 bytes exactly.
ADDED part 3: eui.exe and euiw.exe on the Nov28 build of v4.1.0 work properly.
ADDED part 4 : version 4.0.4 euiw.exe also trucates at 2^32.
useless
2. Re: filesize error
- Posted by CoJaBo2 Dec 01, 2012
- 1886 views
I posted a simplified case (needs no input files) here- http://openeuphoria.org/pastey/174.wc
Ticket here- http://openeuphoria.org/ticket/825.wc
It works under 4.0.4 on 32-bit Linux; looks like a bug specific to Windows version of Eu.
3. Re: filesize error
- Posted by jimcbrown (admin) Dec 01, 2012
- 1849 views
I posted a simplified case (needs no input files) here- http://openeuphoria.org/pastey/174.wc
Ticket here- http://openeuphoria.org/ticket/825.wc
It works under 4.0.4 on 32-bit Linux; looks like a bug specific to Windows version of Eu.
I suspect it won't make a difference, but has anyone tried 4.0.5 ?
Another thing to try would be a MinGW build of 4.0
4. Re: filesize error
- Posted by useless_ Dec 01, 2012
- 1844 views
I posted a simplified case (needs no input files) here- http://openeuphoria.org/pastey/174.wc
Ticket here- http://openeuphoria.org/ticket/825.wc
It works under 4.0.4 on 32-bit Linux; looks like a bug specific to Windows version of Eu.
I suspect it won't make a difference, but has anyone tried 4.0.5 ?
Another thing to try would be a MinGW build of 4.0
I jumped from v4.0.4 to v4.1.0.Nov28 because of the leak fixes in the latter.
However, there's a difference in the way system_exec() is handled, which i gave up working against at 2:45AM.
useless
5. Re: filesize error
- Posted by jimcbrown (admin) Dec 02, 2012
- 1812 views
I posted a simplified case (needs no input files) here- http://openeuphoria.org/pastey/174.wc
Ticket here- http://openeuphoria.org/ticket/825.wc
It works under 4.0.4 on 32-bit Linux; looks like a bug specific to Windows version of Eu.
I suspect it won't make a difference, but has anyone tried 4.0.5 ?
Another thing to try would be a MinGW build of 4.0
I jumped from v4.0.4 to v4.1.0.Nov28 because of the leak fixes in the latter.
I was just looking for additional data points that could narrow the cause of the issue.
6. Re: filesize error
- Posted by CoJaBo2 Dec 02, 2012
- 1784 views
I posted a simplified case (needs no input files) here- http://openeuphoria.org/pastey/174.wc
Ticket here- http://openeuphoria.org/ticket/825.wc
It works under 4.0.4 on 32-bit Linux; looks like a bug specific to Windows version of Eu.
I suspect it won't make a difference, but has anyone tried 4.0.5 ?
Another thing to try would be a MinGW build of 4.0
I jumped from v4.0.4 to v4.1.0.Nov28 because of the leak fixes in the latter.
I was just looking for additional data points that could narrow the cause of the issue.
I updated the ticket-
This now appears to be a result of disk corruption; it works for me in 4.0.4 and 4.0.5, unless_ can't reproduce it either (and it sometimes seems to be creating files of negative sizes).
7. Re: filesize error
- Posted by useless_ Dec 03, 2012
- 1843 views
I posted a simplified case (needs no input files) here- http://openeuphoria.org/pastey/174.wc
Ticket here- http://openeuphoria.org/ticket/825.wc
It works under 4.0.4 on 32-bit Linux; looks like a bug specific to Windows version of Eu.
I suspect it won't make a difference, but has anyone tried 4.0.5 ?
Another thing to try would be a MinGW build of 4.0
I jumped from v4.0.4 to v4.1.0.Nov28 because of the leak fixes in the latter.
I was just looking for additional data points that could narrow the cause of the issue.
I updated the ticket-
This now appears to be a result of disk corruption; it works for me in 4.0.4 and 4.0.5, unless_ can't reproduce it either (and it sometimes seems to be creating files of negative sizes).
The display of the filesize is not causing a filesize. Neither is the bad display of filesize causing a bad filesize. When the filesize is being limited to 2^32, it doesn't matter that is being shown, the filesize is unchangingly stable at 2^32.
The error in display is shown in
http://designerthinking.com/images/NEGSIZE1.JPG
http://designerthinking.com/images/NEGSIZE2.JPG
and is NOT the individual filesize on the left, which is correct, but the folder total, which i have placed a red arrow on. There is pattern of clicks to cause the size display error, if i do not click on a file which has a size over 2^32, then Explorer makes no error in the display. Either way, the individual size on the left is correct.
That said, twice today v4.0.4 eui and euiw failed, and twice today both wrote files over 2^32. I do not know what is going on. Eu v4.1.0.Nov28 has not failed in any filesize test.
useless
8. Re: filesize error
- Posted by CoJaBo2 Dec 03, 2012
- 1872 views
I posted a simplified case (needs no input files) here- http://openeuphoria.org/pastey/174.wc
Ticket here- http://openeuphoria.org/ticket/825.wc
It works under 4.0.4 on 32-bit Linux; looks like a bug specific to Windows version of Eu.
I suspect it won't make a difference, but has anyone tried 4.0.5 ?
Another thing to try would be a MinGW build of 4.0
I jumped from v4.0.4 to v4.1.0.Nov28 because of the leak fixes in the latter.
I was just looking for additional data points that could narrow the cause of the issue.
I updated the ticket-
This now appears to be a result of disk corruption; it works for me in 4.0.4 and 4.0.5, unless_ can't reproduce it either (and it sometimes seems to be creating files of negative sizes).
The display of the filesize is not causing a filesize. Neither is the bad display of filesize causing a bad filesize. When the filesize is being limited to 2^32, it doesn't matter that is being shown, the filesize is unchangingly stable at 2^32.
The error in display is shown in
http://designerthinking.com/images/NEGSIZE1.JPG
http://designerthinking.com/images/NEGSIZE2.JPG
and is NOT the individual filesize on the left, which is correct, but the folder total, which i have placed a red arrow on. There is pattern of clicks to cause the size display error, if i do not click on a file which has a size over 2^32, then Explorer makes no error in the display. Either way, the individual size on the left is correct.
That said, twice today v4.0.4 eui and euiw failed, and twice today both wrote files over 2^32. I do not know what is going on. Eu v4.1.0.Nov28 has not failed in any filesize test.
useless
Are you able to check the disks involved for errors? Ideally, both with scandisk/chkdisk and smartctl. It might also serve to do a memory test.
There is no reason Windows should ever be reporting negative sizes, unless something is severely broken somewhere.
Checking SCM, it does not appear that any of the code involving file output has changed between 4.0.4 and 4.1.0. I.e., there should be no reason whatsoever why 4.1.0 would work and 4.0.4 would not. And, there is certainly no reason why such a problem would occur only some of the time, outside of filesystem curruption and/or hardware failures.
9. Re: filesize error
- Posted by petelomax Dec 04, 2012
- 1741 views
The error in display is shown in
http://designerthinking.com/images/NEGSIZE2.JPG
That looks suspiciously like Windows 98, which has a 4GB file size limit iirc (google just confirmed it: http://www.msfn.org/board/topic/142437-exfat-filesystem-for-win-9xme )
10. Re: filesize error
- Posted by useless_ Dec 04, 2012
- 1733 views
The error in display is shown in
http://designerthinking.com/images/NEGSIZE2.JPG
That looks suspiciously like Windows 98, which has a 4GB file size limit iirc (google just confirmed it: http://www.msfn.org/board/topic/142437-exfat-filesystem-for-win-9xme )
Are you saying i am not running winxp and ntfs?
useless
11. Re: filesize error
- Posted by mattlewis (admin) Dec 04, 2012
- 1704 views
The error in display is shown in
http://designerthinking.com/images/NEGSIZE2.JPG
That looks suspiciously like Windows 98, which has a 4GB file size limit iirc (google just confirmed it: http://www.msfn.org/board/topic/142437-exfat-filesystem-for-win-9xme )
Are you saying i am not running winxp and ntfs?
It seemed like he was wondering if you were, because if you were, he had a good explanation. But, yes, get angry because...I don't know. Just get angry...
Matt
12. Re: filesize error
- Posted by useless_ Dec 05, 2012
- 1678 views
The error in display is shown in
http://designerthinking.com/images/NEGSIZE2.JPG
That looks suspiciously like Windows 98, which has a 4GB file size limit iirc (google just confirmed it: http://www.msfn.org/board/topic/142437-exfat-filesystem-for-win-9xme )
Are you saying i am not running winxp and ntfs?
It seemed like he was wondering if you were, because if you were, he had a good explanation. But, yes, get angry because...I don't know. Just get angry...
Matt
I have found that with most humans, i cannot say something pleasantly and be taken seriously.
Being "nice" is seen as apeasement, as weak, as uncertain of oneself. Like yesterday when i asked in a grocery store where the fruit cake was, the stock clerk insisted 1) i didn't know what i wanted and i meant "angel fruit cake", 2) they had no such thing, 3) if they did it would be in the ready-bake cake mix aisle. So yes, i got angry, i raised my voice to extra loud, i explained to her how absurd her assertions were, she ran to a store manager, who pointed out where it was.
So you tell me, Matt, how can i get Pete to understand and believe i am running winxp and ntfs?
useless
13. Re: filesize error
- Posted by mattlewis (admin) Dec 05, 2012
- 1646 views
I have found that with most humans, i cannot say something pleasantly and be taken seriously.
I'm sorry that this is so for you.
So you tell me, Matt, how can i get Pete to understand and believe i am running winxp and ntfs?
If it were me, I'd probably say, "No, this is WinXP with the classic skin, and so the filesystem is NTFS."
If Pete had been abusive or even mildly snarky, perhaps a response such as you gave would have been justified. If you feel that you must lash out when someone has apparently missed a detail you previously gave and attempts to give good faith advice, then you should probably be prepared to be called out on that.
Not that I don't get frustrated when I have to explain myself multiple times. But it doesn't excuse my bad behavior either.
Matt
14. Re: filesize error
- Posted by useless_ Dec 05, 2012
- 1623 views
I have found that with most humans, i cannot say something pleasantly and be taken seriously.
I'm sorry that this is so for you.
So you tell me, Matt, how can i get Pete to understand and believe i am running winxp and ntfs?
If it were me, I'd probably say, "No, this is WinXP with the classic skin, and so the filesystem is NTFS."
If Pete had been abusive or even mildly snarky, perhaps a response such as you gave would have been justified. If you feel that you must lash out when someone has apparently missed a detail you previously gave and attempts to give good faith advice, then you should probably be prepared to be called out on that.
I felt he was saying i don't have a clue what OS or filesystem i am using, much like the ordeal with the fruit cake at the grocery store.
Since Pete wasn't on irc whe i said the following, i'll copy/paste the the following from irc:
> so i ran chkdsk, it reported no errors, i verify my copy of winxp is legal and up to date on patches, i open exploder and click on each file in order of size, up to and including the 5.11G txt file which Euphoria-4.1.0-Nov28 built, and then i click back down the file list , and all files display properly > then i start up the eu app running Euphoria-4.1.0-Nov28 , all is still fine in exploder, until Eu opens the 5.11G file to append more data, then exploder starts showing negative numbers in the right status bar box again > so like i said, it's a matter of the file being open and active, and not a corruted disk or borked os > either way, doesn't explain the different results with 4.0.4 vs 4.1.0
useless
15. Re: filesize error
- Posted by useless_ Dec 07, 2012
- 1552 views
Whatever the reason(s) i could not do this with Eu v4.0.4, the application i wrote did complete successfully last nite with v4.1.0.Nov28, building a text file of 11.0 GB (11,885,725,664 bytes). So even if there is some error in me being unable to v4.0.4 to run the app over 2^32 bytes, i can confirm that v4.1.0.Nov28 does work as expected.
useless