Difficulty fixing a regression in reading text files with MSVCRT

[Erich Hoover]

In a surprise last-minute request it became necessary for me to get
Athena Visual Studio running under Wine again
(http://appdb.winehq.org/objectManager.php?sClass=version&iId=15183).
A couple of commits to MSVCRT (that are needed for other reasons)
caused regressions in Wine's ability to run this application and I am
having a lot of difficulty getting Athena working again.  The commits
that revealed the problem are:
ebe4a9e321800fbe6d2004a640cd276a28d15c38: Remove CRs earlier in ascii mode.
7f3c70c52fe470d431a51b3363ed4f7336d5f7f5: Remove CRs earlier in ascii
mode in fseek, too.

After tracing down a lot of rabbit holes, it appears that the problem
is somehow related to how the buffer boundary influences the file
position reported by ftell.  I've attached a test case that appears
related to the issue.  I can use a native MSVCRT for now, but I would
appreciate some assistance so that this issue can be properly resolved
at some point.

Erich Hoover
ehoover(a)mines.edu

[Dan Kegel]

Hi Erich,
thanks for that test case.   I'm overloaded at the moment, but
if you file a bug, I'll try to have a look...
- Dan

On Mon, Dec 28, 2009 at 5:03 PM, Erich Hoover <ehoover(a)mines.edu> wrote:
> In a surprise last-minute request it became necessary for me to get
> Athena Visual Studio running under Wine again
> (http://appdb.winehq.org/objectManager.php?sClass=version&iId=15183).
> A couple of commits to MSVCRT (that are needed for other reasons)
> caused regressions in Wine's ability to run this application and I am
> having a lot of difficulty getting Athena working again.  The commits
> that revealed the problem are:
> ebe4a9e321800fbe6d2004a640cd276a28d15c38: Remove CRs earlier in ascii mode.
> 7f3c70c52fe470d431a51b3363ed4f7336d5f7f5: Remove CRs earlier in ascii
> mode in fseek, too.
>
> After tracing down a lot of rabbit holes, it appears that the problem
> is somehow related to how the buffer boundary influences the file
> position reported by ftell.  I've attached a test case that appears
> related to the issue.  I can use a native MSVCRT for now, but I would
> appreciate some assistance so that this issue can be properly resolved
> at some point.
>
> Erich Hoover
> ehoover(a)mines.edu