Hi,

I'm trying to read a file line by line, and I get a core.exception.InvalidMemoryOperationError@(0), even after reducing the program to:

import std.stdio;

void main()
{
  File f = File("testfile");
  foreach(line; f.byLine)
  {
  }
}

The file is a simple table of ascii characters, 811 columns and it fails on the second line. Taking any subset of the columns and the program runs fine so I don't think it can by any particular file corruption.

In this simple example I can get it to work by changing:

foreach(line; f.byLine)

to

foreach(char[] line; f.byLine)

but in my more complicated program this still fails with the same error:

foreach (char[] lineVar; inFile.byLine)
{
  lineVar.split.indexed(places).joiner("\t").writeln;
}

(as does the range version I originally wrote:
inFile.byLine.map!(a => a.split.indexed(places).joiner("\t")).joiner("\n").writeln;)

Is this a bug, gdc on version 2.066 seems to have no problems with it? I'd be happy to fill in a bug report, but I can't share the file as it's sensitive genetic data and I haven't been able to reduce it to anything innocuous.

This has me very puzzled, any suggestions would be much appreciated.

Thanks very much

Andrew

On Tuesday, 15 September 2015 at 13:56:37 UTC, Andrwe Brown wrote:
> I'm trying to read a file line by line, and I get a core.exception.InvalidMemoryOperationError@(0), even after

https://issues.dlang.org/show_bug.cgi?id=13856

Try DMD 2.068, it has got fixed byLine implementation.

On Tue, 15 Sep 2015 13:56:36 +0000
Andrwe Brown via Digitalmars-d-learn
<digitalmars-d-learn@puremagic.com> wrote:

> Hi,
> 
> I'm trying to read a file line by line, and I get a core.exception.InvalidMemoryOperationError@(0), even after reducing the program to:
> 
> import std.stdio;
> 
> void main()
> {
>    File f = File("testfile");
>    foreach(line; f.byLine)
>    {
>    }
> }
> 
> The file is a simple table of ascii characters, 811 columns and it fails on the second line. Taking any subset of the columns and the program runs fine so I don't think it can by any particular file corruption.
> 
> In this simple example I can get it to work by changing:
> 
> foreach(line; f.byLine)
> 
> to
> 
> foreach(char[] line; f.byLine)
> 
> but in my more complicated program this still fails with the same error:
> 
> foreach (char[] lineVar; inFile.byLine)
> {
>    lineVar.split.indexed(places).joiner("\t").writeln;
> }
> 
> (as does the range version I originally wrote:
> inFile.byLine.map!(a =>
> a.split.indexed(places).joiner("\t")).joiner("\n").writeln;)
> 
> Is this a bug, gdc on version 2.066 seems to have no problems with it? I'd be happy to fill in a bug report, but I can't share the file as it's sensitive genetic data and I haven't been able to reduce it to anything innocuous.
> 
> This has me very puzzled, any suggestions would be much appreciated.
> 
> Thanks very much
> 
> Andrew

Which OS?

On Tuesday, 15 September 2015 at 14:19:13 UTC, Daniel Kozák wrote:
>
> Which OS?

It's CentOS release 6.5 (Final), I tried dmd 2.068.1 and the problem has disappeared. Thanks very much for the advice, I can stick to old gdc for speed until ldc catches up to 2.068.

Best

Andrew

For reference, it was this PR:
https://github.com/D-Programming-Language/phobos/pull/3089
which fixed the same issue for me.

On Tuesday, 15 September 2015 at 14:55:42 UTC, Martin Krejcirik wrote:
> For reference, it was this PR:
> https://github.com/D-Programming-Language/phobos/pull/3089
> which fixed the same issue for me.

A very naive question: would it be possible in this case to backport it into gdc/ldc by copying the pull request and building the compiler from source, or would this get me into a world of pain?

On Tuesday, 15 September 2015 at 15:28:23 UTC, Andrew Brown wrote:
> A very naive question: would it be possible in this case to backport it into gdc/ldc by copying the pull request and building the compiler from source, or would this get me into a world of pain?

Cherry-picking should work and merge cleanly. I have done it for DMD 2.067. I don't know how difficult it is to recompile Phobos and Druntime with LDC/GDC though.

Thanks very much for your help, it seemed to work a treat (I hope :))! Compiling ldc wasn't too bad, make the changes to runtime/phobos/std/stdio.d and then just building as normal was no problem. Unittests are passing and it handles that file perfectly.

On Tuesday, 15 September 2015 at 16:11:06 UTC, Martin Krejcirik wrote:
> On Tuesday, 15 September 2015 at 15:28:23 UTC, Andrew Brown wrote:
>> A very naive question: would it be possible in this case to backport it into gdc/ldc by copying the pull request and building the compiler from source, or would this get me into a world of pain?
>
> Cherry-picking should work and merge cleanly. I have done it for DMD 2.067. I don't know how difficult it is to recompile Phobos and Druntime with LDC/GDC though.