editline truncates at 64 characters in batch mode

[Mark O'Donohue]

In batch mode, if the input line is greater than 64 characters, the first 64 characters are lost.

In my case:

$isql-fb < test.sql

the input file of:

SET ECHO ON;
SELECT
  dt.*
FROM
  Table_10 t10
  XXXXJOIN (SELECT * FROM Table_10 t2 WHERE t2.ID = t10.ID) dt ON (1 = 1);

becomes:

SELECT
  dt.*
FROM
  Table_10 t10
N (1 = 1);

In the above the line :
  XXXXJOIN (SELECT * FROM Table_10 t2 WHERE t2.ID = t10.ID) dt ON (1 = 1);

became:
N (1 = 1);

This does not happen in interactive mode, nor does it happen with the version of editline that is included in the firebird source distribution.

This also seems to have been reported and fixed for minix-editline here:

           http://vmlinux.org/jocke/blog/hacks/

I am still investigating, but this first came to our notice with trying to run the python QMTest tests on the Firebird package distributed with Ubuntu.

Editline version is the latest: 1.12-5 and a quick look on the Debian package, didnt show any latter changes or bug reports.

General info:

$ lsb_release -rd
Description: Ubuntu 7.10
Release: 7.10

$ apt-cache policy libeditline0
libeditline0:
  Installed: 1.12-5
  Candidate: 1.12-5
  Version table:
 *** 1.12-5 0
        500 http://au.archive.ubuntu.com gutsy/universe Packages
        100 /var/lib/dpkg/status

Cheers - Mark

[Mark O'Donohue (mark-odonohue)]

Ok here is a simple program showing the problem.

It is actually the testit.c program distributed with libeditline,

$make
$./run.sh

essentiallty it does

$./testit < test.txt

Where test.txt has lines over 64 character long.

Cheers - Mark

[Mark O'Donohue (mark-odonohue)]

The buffer in editline.c
      STATIC char* read_redirected()

is 64 characters long, when it get to the end of 64 characters it does:

        if (p == end) {
            size += MEM_INC;
            p = line = realloc(line, size);
            end = p + size;
        }

So 'p' which is where it is writing the next character is set back to the start of the line, not left at the end. The attached patch, bumps p to next write position correctly.

        if (p == end) {
            int lengthSoFar = p - line;
            size += MEM_INC;
            line = realloc(line, size);
            p = line + lengthSoFar;
            end = line + size;
        }

Cheers - Mark

[Brian Murray (brian-murray)]

Looking at the attachments in this bug report, I noticed that "fix for bug, based on debian editline 1.12 base" was not flagged as a patch. A patch contains changes to an Ubuntu package that will resolve a bug and this attachment is one! Subsequently, I've checked the patch flag for it. In the future when submitting patches please use the patch checkbox as there are some Launchpad searches that use this feature. Thanks for your contribution Mark O'Donohue!