From: Gabriel Gerhardsson (mr.g@home.se)
Date: Tue May 21 2002 - 18:34:49 EDT
Hi Jeremy
It's very educating to watch how someone who knows the Abi source, among
other things, solve some of the problems I had. I looked at your code
and found it very nice, indeed. As you say your code is almost XP and
obviously much better in a number of places, than the old solutions.
Unfortunaly I won't have time to work on Abi for another two weeks or
so. After that, I'll have some more time to adapt your changes into XP
land and such, if you don't beet me to it that is :) Feel free to...
Btw, one of the things I want to do is to separate out the download part
and make a nicer dialog with a Cancel button. Current behaviour isn't
very friendly to the user... there's no way to stop the download but to
unplug the network interface or kill Abi. Not nice.
But as usual, we'll see how much spare time there will be.
Take care
/Gabriel
On Tue, 2002-05-21 at 23:21, Kenneth J.Davis wrote:
> CVS: ----------------------------------------------------------------------
> CVS: Enter Log. Lines beginning with `CVS:' are removed automatically
> CVS:
> CVS: Committing in .
> CVS:
> CVS: Added Files:
> CVS: ap_Win32HashDownloader.cpp ap_Win32HashDownloader.h
> CVS: ----------------------------------------------------------------------
> initial Win32 specific implementation for automatic dictionary downloader
>
> This is the platform specific portion, I still need to commit
> the patches to the xp files so they successfully build on Windows and
> the patches to the build system to allow building with this enabled.
>
> I wanted to see how this works, and since I don't use the Unix versions
> went ahead and did the tiny Win32 specific part. Pretty cool! :-)
>
> This implementation should be nearly completely xp, instead of spawning programs
> it downloads the .tgz dictionary files and based on untgz.c in zlib contrib
> directory it makes appropriate zlib calls to extract out the files and
> place in the proper place. (I don't believe there are any license problems
> with the small amount of code used from untgz.c, but if anyone disagrees
> please let me know.) The rest of the code is based on the Unix version.
> Only the appropriate dictionary hash file and its encoding are extracted,
> all other aspects of the tarball are ignored, including any paths prefixing
> the hash file and encoding.
>
> I still need to double check the xp code for assumptions, off hand I think
> everything is ok except the implicit use of /tmp directory. Also I would
> like to allow the profile to be able to override the hard coded URLs
> (e.g. so someone using this within a firewall could get the files from an internal server).
This archive was generated by hypermail 2.1.4 : Tue May 21 2002 - 18:40:57 EDT