OT: Looking for a Utility

I'd guess there are lots of single bit errors that never get as far as a crash.

I wonder whether it would be possible to find a board with error checking, fiddle with the setup to just count the errors but not correct them, and let Windows run.

There are "bad sticks", with the handling some people do I'm not surpised, but I don't think those are the same as the rare single bit failure.

There are also bad contacts. Someone inside Intel told me that he had found out that there was a screwup in the standarization process. Somehow the guys writing the specs for the edges of the dimms and the guys writing the specs for the sockets got out of sync. The result was that number of times you can insert and remove a dimm and still expect it to not have any errors is substantially less than it was supposed to be. I asked him how many times he thought was safe and he said "half a dozen, maybe."

If you dumped 'doze and ran OS/2 or eCS, you could type that line into an OS/2 command line and hit enter right from the desktop. You could even shorten up the process using ZTree. OS/2 runs that stuff with far less fuss than 'doze.

Ted

I haven't had a board with ECC for a while, but as you say, the error rate isn't real high. Back in the days of 256K ram chips, small amounts of radioactive material that naturally occurred in the ceramic almost shut semiconductor memory down. There are still errors caused by cosmic rays, and other ionizing radiation that strikes a memory cell from time to time.

The problem is, that without error detection, you have no way of telling when, or if, single bit problems are occuring.

In the case of my last bad stick, the error tended to happen in a critical part of memory, so it would crash the kernel in linux... sometimes. Perhaps once every month. A completely unacceptable rate for linux, or any other unix. I swapped the old sticks out for new, and the crashes went away. Later, after doing extensive memory checks on the bad sticks, and finding nothing wrong, I put them in another machine, and the monthly crash moved to that machine. The next step was the trash can.

I haven't seen any SIMMS, or other memory sticks that looked damaged from the sockets, but I am sure that that is a possibility.

-Chuck

I think it would be pretty easy to write a program which reads a list of input files, and writes them all, in the order they are listed, into an output file.

It could be called mycat.

e.g.:

mycat listfile.txt outfile.ps

listfile.txt would look like this:

preamblefile file1 file2 file3 file4 lastfile

I could do this for you for nothing, if it would solve your problem adequately.

You could also install cygwin. Cygwin comes with bash, which would let you use cat, and make composing an error free command line trivial.

--Mac

Some computers still use ECC RAM. A few years ago I designed a SBC with ECC RAM. One of the chipset registers could set the computer to raise a system error signal, thus blue-screening windows 2000 upon detection of a single bit error. The BIOS allowed you to set this register. We exercised these SBC's heavily: over temperature, over Voltage, and over clock rate. Once we got the thing dialed in, we NEVER saw memory errors.

Perhaps your comments apply to crappy computers, but well-designed ones should not exhibit these kinds of problems.

By the way, Windows NT is posix compliant. Posix != unix.

--Mac

Oh come on it's great fun. You need to keep your hand in !

See my post elsewhere about using debug to do a low level format. Just needs 2 lines to be typed in. One of which is debug < enter >.

God knows what it would do now. I'm not risking it. Probably still works.

Graham

Why don't you create a small .cmd or .bat file that does the copy for a single file? Just append a file (copy total.ps to tmp.ps, then the new+tmp to total again) The new file (%1) can be given to the command file by dropping the fileX.ps onto it. Assuming drag and drop is fine for you. Still one file a time, but not typing anymore.

Cheers,

Joop

"Jim Thompson" ...

If the files were generated in the proper sequence into an empty directory (not unlikely) then running:

copy /b %thatdir%\\*.ps total.ps

just below that directory should do the trick, since it appends all files in the order in which they are in the directory. Be sure to use a DOS version that understand long filenames (WIN2K+ at least works).

You could use a batch file with a line like this, and add it with a nice name to the function list for handling directories. Open Explorer, then open menus:

Explorer -->

Tools Folder options File Types File Folder Advandced New

Then a right-click on a directory shows your option to merge all it's *.ps files.

Arie de Muynck

I'd heard of doing this with PS files but couldn't get it to work. Are there special requirements on the ps files?

Ted

Of course! But comsumer grade computers, as made by Dell, HP, and others are primarily cost driven; as a result, non parity, non ECC memory gets designed into the systems. You and I have no way of knowing for sure how many single bit errors simply happen undetected in these systems. A system that uses non error detecting memory is not well very designed.

Most of the memory I have used in recent years is amazingly reliable, I think (without error detection hardware, I cannot know). But I have come across some that after of days of testing, using some very sophisticated memory test programs, passes all of the testing with flying colors! But when used to run linux, fails in a sporadic mannor. Other identical memory sticks don't fail. Obviously something was happening that the memory test program was ill equiped to find. But then, I have never met a memory test program that could successfully find anything but the most drastic of memory problems. ...

I am aware of that. Posix is simply an IEEE committee's attempt to set a standard set of interfaces, libraries, and behaviors. The Posix committee used unix as a model for most of what they did. One of the first goals set within the linux/gnu community was to become fully posix compliant.

-Chuck

It uses the order the files appear in under W.E. So, if you have the folder set to display files alphabetically by filename, then that's what Concat will use. However, Concat also shows you a list of the filenames in the selected order, and you can manually move them around (I think this manual re-order may require registration, but there are serials around I think).

Anyway, with this available, I see no reason to use Copy, or to reinvent the wheel. It works exactly the way you want it to, it's fast, and it's in the context menu.

That works nicely. Thanks for the pointer. I should have thought to look there myself as I have some other utilities by JDDesign.

...Jim Thompson

Thanks, Jim. Saved for later use.

Ted

high light the files to be catted. Each highlighted file

file. :"

ConCat from JDDesigns does it perfectly, and allows setting order of concatenation.

...Jim Thompson

late reply here.. But

You can create a batch file and add it to the 'Send to' folder. In explorer high light the files to be catted. Each highlighted file name then will be sent as an argument to the batch file.

So then the batch file might look like this.....

@echo off SET /P cat_filename="Please enter a (path) & filename for the concatanated file. :" echo %1 %2 %3 %4 :BEGIN if '%1'=="" goto END copy %1 /A temp.txt /A shift :MYLOOP echo %1 copy /A temp.txt /A + %1 /A %cat_filename% /A shift copy %cat_filename% /A temp.txt /A if '%1'=='' goto END goto MYLOOP echo done... :END del temp.txt pause

One problem is that it does not sort the input files ;(

Martin

high light the files to be catted. Each highlighted file

file. :"

I checked my bat file again and it can sort, the file names are passed in the order as shown in explorer. Right clicking on the first file will order it from the top. much like ConCat. But thats about the limit.

Cheers

Hi Jim,

Funny how things go. When you first posted this trick, I had absolutely no need to concatenate postscript files. Then, a couple of days later, a very urgent need suddenly materialized!

I tried your trick and several others, but they would not work on my system. In case others also run into problems, here's another method recommended by Adobe. They mention this method may help reduce the overall file size by merging the font tables into a single table, instead of having multiple copies when files are simply concatenated.

This method uses a postscript file to call a routine in Adobe Distiller. One the routine is defined, the list of desired postscript files is added and the file is saved as a .ps

You can put the pages in any desired order. For example, I found sorting them in reverse order greatly simplified printing.

Here's the Distiller routine:

---------------------------------------------------------------------

%! % PostScript program for distilling and combining multiple PostScript % files. When embedding font subsets, it is highly recommended you use % this technique to distill multiple PS files so only one font subset % is used for each font.

/prun { /mysave save def % Save before running this file dup = flush % Shows name of PS file being run RunFile % Calls built-in Distiller procedure clear cleardictstack % Cleans up after PS file mysave restore % Restores save level } def

% Here's the list of files to concatenate. I found the full path is % needed. Note the double backslash used as a separator.

(f:\\\\0pera\\\\cache\\\\page01.ps) prun (f:\\\\0pera\\\\cache\\\\page02.ps) prun (f:\\\\0pera\\\\cache\\\\page03.ps) prun (f:\\\\0pera\\\\cache\\\\page04.ps) prun

% INSTRUCTIONS % % 1. Locate all PostScript files to be distilled. % % 2. Make a copy of this file and give it the name you want to have as % the prefix for the resulting file. For example, you could name this % file MyBook.ps. % % 3. Include a line for each PostScript file to be run using the % pathname syntax appropriate for the platform running Acrobat % Distiller: % % Macintosh pathname syntax: (File.ps) prun % Windows pathname syntax: (c:\\\\mydir\\\\file.ps) prun % % 4. Load this file in Acrobat Distiller. It will automatically % execute and produce a pdf with the same prefix.

---------------------------------------------------------------------

Mike Monett