iceman_k
18th February 2003 23:39 UTC

NSISdl bug?
I tried using NSISdl with an invalid URL to see what it would do.
The error message that is returned has unprintable ASCII characters in it.
It's as if it is pointing to freed or uninitialized memory.
This is the code I wrote:

NSISdl::download /TIMEOUT=30000 "http://ardownload.adobe.com/pub/adobe/acrobatreader/win/5.x/5.1/AcroReader51_ENU.exe1" "$TEMP\AcroReader51_ENU.exe"
StrCmp "$0" "Success" done
DetailPrint "$0"
done:

The URL is incorrect, it should be:
"http://ardownload.adobe.com/pub/adobe/acrobatreader/win/5.x/5.1/AcroReader51_ENU.exe"
With the correct URL, the plugin works great.
However, shouldn't it have a meaningful error message for bad URLs?
---
Thanks

Joost Verburg
19th February 2003 13:05 UTC

The current CVS version of NSISdl does not write the return value to $0, but pushes it to the stack, like other plugins.

iceman_k
19th February 2003 14:56 UTC

That explains it. However, the Readme.txt needs to be updated to reflect this.
---
Thanks

Joost Verburg
19th February 2003 15:00 UTC

The Readme has already been updated. You probably got this info from the old Readme file.

iceman_k
20th February 2003 01:35 UTC

Is this in the 2.0b1 release or the latest dev snapshot?

kichik
20th February 2003 05:03 UTC

Latest development snapshot (CVS).