|
From: | Peram, Sudhakara |
Subject: | RE: ci failed for large files (>2GB) - diff: memory exhausted |
Date: | Thu, 20 Jan 2011 10:40:00 -0500 |
Hi All I found interesting thing with
diff command, I have compiled diff with 64-bit to support large files (>
2GB) and it is working fine without using –a option. When I use –a
option on large files I am getting below error. Posted a question to diff
command group as well, but no response. Error Message: diff: memory exhausted ‘–a’ option
used to “Treat all files as text and compare them line-by-line, even if
they do not seem to be text”. We have built RCS with GNU diff
utility and with ‘–n’ and ‘–a’ options for
diff, so check-in is failing for large files. Here I have doubt whether RCS
will work without ‘–a’ option or not. Can anyone tell me if you
already tried without ‘–a’ option. My server architecture is : SunOS hostname 5.9 Generic_122300-47
sun4u sparc SUNW,Ultra-4 Thank you in Advance! --Sudhakar -----Original Message----- Thanks Hugh for your response! Yeah I know that RCS will not be able to produce the
differences (i.e., rcsdiff) between two versions of binary file. As per my
understanding ci will use the diff command to check whether the latest
file and the last version of that file are being same or different before
check-in. And for the file less than 1GB it is working fine. I have
tried to check-in the file with 5.5GB and it worked fine for the initial
revision. Let me see if someone can help me in this regard. --Sudhakar -----Original Message----- From: Hugh Sasse [mailto:address@hidden Sent: Thursday, January 13, 2011 2:15 PM To: Peram, Sudhakara Cc: address@hidden Subject: RE: ci failed for large files (>2GB) - diff:
memory exhausted i can't remember for certain, but I am pretty sure that
RCS does not support binary files. I think the newer systems
like git will handle this sort of thing, but it seems that differences are not
really done for binary files, at least going by casual observation of the
size and description of changes. If you need the compression afforded by
differences for binary files, then I'm not able to make any suggestions. Hopefully someone else who knows more can respond. Hugh On Thu, 13 Jan 2011, Peram, Sudhakara wrote: > Hi Hugh > > I have re-compiled diff program with large file
support and it is also > 64-bit compilation. Here one more thing diff command
executed > successfully when I ran it manually on the
same set of files which were > used for ci command. > > Note: These files are binary in nature, not the text
files. > > Thanks for your help! > > > --Sudhakar > > -----Original Message----- > From: Hugh Sasse [mailto:address@hidden > Sent: Thursday, January 13, 2011 1:06 PM > To: Peram, Sudhakara > Cc: address@hidden > Subject: Re: ci failed for large files (>2GB) -
diff: memory exhausted > > Just wondering if you compiled the diff program with
large file support? > > Hugh > > On Wed, 12 Jan 2011, Peram, Sudhakara wrote: > > > Hi All > > > > > > > > I have compiled RCS with 64-bit to support
large files (> 2GB) and it > is > > working fine for initial revision. But from
second revision onwards it > > is failing if the file size >= 1.1 GB and I
am getting the following > > error. This error is more related to diff
utility which is used by ci > > command while check-in the file into RCS. > > > > > > > > Error Message: > > > > diff: memory exhausted > > > > > > > > I did google for this error and I came to know
that there some other > > utilities to get rid of this error, but here I
am not issuing diff > > command manually (i.e., diff command is issued
by ci command > internally) > > so I can't use the other available
utilities/options. Is there anyway > > get rid of this error like suppressing diff
command while check-in or > > compiling diff utility with large file support? > > > > > > > > My server architecture is : SunOS hostname 5.9
Generic_122300-47 sun4u > > sparc SUNW,Ultra-4 > > > > > > > > Thanks in advance for your help! > > > > > > > > > > > > Regards > > > > Sudhakar > > > > > > > > > |
[Prev in Thread] | Current Thread | [Next in Thread] |