NSIS 3.0a1 — NSIS Forum Archive

TrifonovSNovember 13, 2013 at 6:39 AM#

When I compile my existing projects with the new NSIS version 3.0a1, it prints only several lines in the output window of the editor. With the older version 2.46 this is different. I see a lot of prints during the whole compilation. Of course I don't read all of them, but this can give me information what the compiler is doing currently. My script is huge and it needs probably 7-8 minutes for compilation and without prints with V3.0a1, it looks that it hangs and I don't know when it will finish. So is this behavior expected? Can I enable the prints similar to V2.46?

AndersNovember 14, 2013 at 12:45 AM#

Originally Posted by TrifonovS View Post

When I compile my existing projects with the new NSIS version 3.0a1, it prints only several lines in the output window of the editor. With the older version 2.46 this is different.

This change is listed in the release notes. It should also compile a little faster without all the prints.

Use makensis -v4 foo.nsi to print everything...

TrifonovSNovember 21, 2013 at 6:06 AM#

I think that there is a problem with some existing plug-ins when the unicode support is enabled (Unicode true). For example Aero plug-in can not show correct some German letters. The problem is not in my code, because when I don't enable Aero plug-in the branding text is shown correct. I also had a problem with the caption-text of the inetc plug-in. When I disable unicode support all works fine. But when I enable it, the installer stops with a strange message.

AndersNovember 21, 2013 at 7:49 AM#

Originally Posted by TrifonovS View Post

I think that there is a problem with some existing plug-ins when the unicode support is enabled (Unicode true). For example Aero plug-in can not show correct some German letters. The problem is not in my code, because when I don't enable Aero plug-in the branding text is shown correct. I also had a problem with the caption-text of the inetc plug-in. When I disable unicode support all works fine. But when I enable it, the installer stops with a strange message.

I thought Afrow UK already said this was a bug in his plugin?

For anything else, post some example code or something, we are not mind readers...

TrifonovSNovember 21, 2013 at 9:56 AM#

I thought Afrow UK already said this was a bug in his plugin?

This was for showing of the ampersand symbol. I though that this is different problem.

For anything else, post some example code or something, we are not mind readers...

Of course, you are right. I was under a big time pressure with my project and I came back to NSIS V2.46. I will try to find time to install again NSIS V3.0a1 and to prepare simple code that can reproduce the problem (at least this one with inetc plug-in). I just though that might be useful to write some notes, because probably somebody already knows about these problems.

TrifonovSNovember 21, 2013 at 12:42 PM#

So here is simple script that shows a strange behavior of the inetc plug-in. It just tries to download SP3 for Windows XP from a predefined location. When Unicode is true, after execution, it doesn't download anything, but just exits and the returned value is shown in a message box. The return value is unreadable. But when Unicode is set to false, the downloading is started. At the end (or if it is interrupted) the returned value is readable as it is expected.

AndersNovember 21, 2013 at 1:16 PM#

Originally Posted by TrifonovS View Post

So here is simple script that shows a strange behavior of the inetc plug-in. It just tries to download SP3 for Windows XP from a predefined location. When Unicode is true, after execution, it doesn't download anything, but just exits and the returned value is shown in a message box. The return value is unreadable. But when Unicode is set to false, the downloading is started. At the end (or if it is interrupted) the returned value is readable as it is expected.

...and you are using the Unicode version of the plugin?

AndersNovember 21, 2013 at 1:41 PM#

When strings are wrong/corrupted/1 character long you should always make sure that you are using the correct plugin that matches your target.

*Unicode true
*!addplugindir to inetc unicode folder
*change $TEMP\ServicePack3 to $TEMP\Download in inetc::get call

and it works...

TrifonovSNovember 21, 2013 at 4:28 PM#

That's right! I really forgot that I have to use an Unicode version of the plug-in. Thanks a lot and sorry for my mistake.

roytam1November 27, 2013 at 12:00 AM#

does FileRead and FileReadByte broken if Unicode true?

AndersNovember 27, 2013 at 7:10 AM#

Originally Posted by roytam1 View Post

does FileRead and FileReadByte broken if Unicode true?

Fixed in next alpha:

FileRead in Unicode installers can handle DBCS, conversion output is limited to UCS-2.

FileRead in Unicode installers now uses the Unicode replacement character (U+FFFD) for invalid characters and not '?'.

FileReadByte no longer performs a Unicode conversion on non-ASCII characters