Re: Reading and writing a big file in Ada (GNAT) on Windows XP

[Fionn Mac Cumhaill <invisible@hiding.from.spam>]

On 2 May 2007 00:46:07 -0700, george@gentoo.org wrote:

>Fionn Mac Cumhaill wrote:
>> All it does is read lines in a loop from a text file with
>> Ada.Text_IO.Get_Line, does minor modifications on about 80% of the
>> lines that it reads, and writes the lines to an output file with
>> Put_Line.
>Looks like a perfect task for sed utility to me. Even if sed does not
>do all you need you can at the very least check some simple sed
>transformation on your data, which should give you an estimate of what
>run times you can expect with your file.
>I am not familiar with MinGW environment. If it is anything like
>CygWin then you should be able to run GNU tools and perform that test
>with sed. At least this should answer the
>
>> Do I have an Ada problem, a GNAT problem, or a MinGW problem?
>
>(and CygWin does tend to be slower than native computation, however I
>used it only a few times).
>
>
>George
I considered sed, but I know next to nothing about it, and decided
that by the time I learned enough to use it, I could have done the job
in Ada several times over.

I did do further investigation; I made a copy of the now-working
program and threw most of the program away, leaving only a very simple
program which read the large input file, but made no changes and did
no output. I added code to track the run time and put the buffer clear
back in. It read the 10 million lines in just a little over five
minutes. I then put  Index back and used it to search the buffer for a
short string that would never be found, seatching forwards from the
beginning of the input buffer. Bingo. Run time increased to a bit more
than 1-1/2 hours. 

One of the earlier commenters observed that there was a bug in GNAT's
Index. It will be interesting to try this again once the bug is fixed.