Re: proposed patch to mktexlsr

Olaf Weber <[email protected]> schrieb:

> Thomas Esser writes:
>
>>> - if we don't have write permissions to the directory, but the file
>>>   exists yet, we don't skip this directory. Below the script checks if
>>>   the file is writeable, anyway:
>
> The way we handle the file update requires write permission to the
> directory as well as the file itself.  I fail to see why this change
> is an improvement.

You're right. I missed that this change is in the context of a further
change. We also apply the following patch:

-  # To be really safe, a loop.
-  until PERMS=`kpsestat = "$db_file"`; do sleep 1; done
-  chmod $PERMS "$db_file_tmp"
-  rm -f "$db_file"
-  mv "$db_file_tmp" "$db_file"
-  rm -rf "$db_dir_tmp"
+  cat "$db_file_tmp" > "$db_file"
+  rm -f "$db_file_tmp"

But db_file_tmp is created securely by our tempfile program. 

I am not sure whether this has any other implications (and I didn't
develop it). Just for your information I attach the complete
Debian-specific patch that we apply to mktexlsr, there's also hunks for
other files in it.

Regards, Frank


-- 
Frank Küster, Biozentrum der Univ. Basel
Abt. Biophysikalische Chemie