- NSIS Discussion
- the future of UAC?
Archive: the future of UAC?
Yathosho
7th November 2011 22:49 UTC
the future of UAC?
(please don't read this as if it was a rant)
1. the 0.2.3 branch in only available for ANSI, while at the same time it seems the Unicode branch of NSIS that appears to have realistic future
2. anders offers an experimental build through IRC, but all the servers mentioned in the FAQ and on the NSIS Wiki are either down or not accessible for the public ("Permission denied")
3. the syntax has changed in 0.2.3 and the documentation is bad if present at all (is there an equivalent to UAC::ExecCodeSegment ?)
i'd be perfectly happy with 0.2.2 if it wasn't for the glitch in the load screen. while at it, let me also mention that the load screen of 0.2.3 is terrible. placing the icon is the middle is against all rules gui design, i also think that it would be fair to the user to include a status message with that dialog.
also what are the plans for the plugin? not releasing it for unicode is not optimal, but i understand if the author isn't happy about maintaining two versions. still i'd like to know more on that decision and drive away the "discussion" from the talk page (where hardly anyone is looking)
Anders
8th November 2011 00:46 UTC
(I love rants)
1. There is no official unicode build, unless someone steps up you will never see an official unicode build. I don't think it is a good idea to support the fork so I will not provide builds for it ATM
2. I don't control the internet (sorry), but the download on the main wiki page should work.
3. Did you look in UAC.nsh? Anyway, unless you are calling UAC_AsUser_ExecShell on the finish page you are probably doing something the plugin was not designed to do AKA the basic example on the current wiki page is the intended usage of the plugin. The plugin was never designed to be user friendly, it was really just a experiment in the early days of Vista.
The load screen is only there to deal with loss of focus in .oninit and the icon was the best I could do (What do you suggest the status message should say?)
There are no future plans for the plugin, IMHO it is more important that we get a new NSIS release and I am working on that... (If someone wanted to take over the plugin it would probably be a better idea to start over from scratch since the source code is a ugly mess full of hacks)
Yathosho
22nd November 2011 09:59 UTC
slowly getting back after refurnishing my flat, sorry for the late reply
Originally posted by Anders
1. There is no official unicode build, unless someone steps up you will never see an official unicode build. I don't think it is a good idea to support the fork so I will not provide builds for it ATM
i know unicode isn't official, but unlike the official build it makes the impression of progressing. despite svn commits, there haven't been new versions in two years - and isn't that against the open source mantra of "release often, release early"? even a service release with updated plugins would be better than no release and would send a sign to developers that are uncertain about nsis' future and tend to switch to the unicode build. anyway, this is to be discussed elsewhere.
2. I don't control the internet (sorry), but the download on the main wiki page should work
i was only referring to the experimental build, which you
might give away via irc only. thanks to the wiki's history, one can access some older versions of the uac plugins and despite an annoying visual glitch one older version is doing the job for me (and it comes in ansi and unicode flavors)
i'm not always happy with plugin maintenance in general. i'd like to see some guidelines for developers that suggest certain ways of how to publish a plugin (e.g. use of wiki pages, version history including downloads for old versions, licensing, availability of source code, documentation standards)
3. Did you look in UAC.nsh?
i did due to the lack of documentation. needless to say this isn't ideal, which is why i'd like to see old versions of the plugin still being available for download (without digging through the wiki history) and a newer version without documentation marked experimental/beta/etc.
The load screen is only there to deal with loss of focus in .oninit and the icon was the best I could do (What do you suggest the status message should say?)
in a sense i don't really care what it says, as i probably wouldn't have the time to read it. i don't how this is different on slow computers, but if the banner appears there should be some feedback on what is going on. so why not just put a status message saying what's actually happening (e.g. "executing setup with elevated rights")
Anders
22nd November 2011 12:03 UTC
Originally posted by Yathosho
i know unicode isn't official, but unlike the official build it makes the impression of progressing. despite svn commits, there haven't been new versions in two years - and isn't that against the open source mantra of "release often, release early"? even a service release with updated plugins would be better than no release and would send a sign to developers that are uncertain about nsis' future and tend to switch to the unicode build. anyway, this is to be discussed elsewhere.
There will be a 2.47, alpha by xmas if we are lucky but there is no rush and no specific target date. Which new feature from SVN is so important that you need a release right now? Even if we wanted to do a release today we can't since there are at least two show stopping things that need to be fixed before a we can do a release. If you need a sign so bad, check the log, I am committing fixes and features almost every day.
Originally posted by Yathosho
i was only referring to the experimental build, which you might give away via irc only. thanks to the wiki's history, one can access some older versions of the uac plugins and despite an annoying visual glitch one older version is doing the job for me (and it comes in ansi and unicode flavors)
There is no might, I believe I have given it out to everyone that asked on IRC. I have not tested the plugin, do you think I should post a totally untested plugin on the wiki? I'd rather work on 2.47, testing and debugging a plugin on a unsupported platform will have to wait...
Originally posted by Yathosho
i'm not always happy with plugin maintenance in general. i'd like to see some guidelines for developers that suggest certain ways of how to publish a plugin (e.g. use of wiki pages, version history including downloads for old versions, licensing, availability of source code, documentation standards)
Everything is zlib unless stated otherwise. I tried to come up with a standard wiki header for plugins but it does not seem like anyone is following my lead but it is a wiki so if you are unhappy there is nothing stopping YOU from fixing things...
Originally posted by Yathosho
i did due to the lack of documentation. needless to say this isn't ideal, which is why i'd like to see old versions of the plugin still being available for download (without digging through the wiki history) and a newer version without documentation marked experimental/beta/etc.
The syntax changed after I wrote that html readme so it is no longer part of the docs (It used to document the exact behavior of the export functions but this really limits future changes and features). The code example on the wiki shows the intended usage, if you feel more documentation is needed you can edit the wiki or send me a suggestion of what you think the docs should look like...
Originally posted by Yathosho
in a sense i don't really care what it says, as i probably wouldn't have the time to read it. i don't how this is different on slow computers, but if the banner appears there should be some feedback on what is going on. so why not just put a status message saying what's actually happening (e.g. "executing setup with elevated rights")
In most cases you are on the secure uac desktop so you can't really read any message, the uac prompt itself is a indicator of what is happening.
Yathosho
22nd November 2011 12:29 UTC
There is no might, I believe I have given it out to everyone that asked on IRC
let's make this short (and go back to my initial post). that's exactly what i tried in the first place and that's also what lead to my initial post. the confusion starts with the correct irc server, the wiki mentions one, the forum another - both didn't work (one timed out, the other stated i had no authorization to connect)
In most cases you are on the secure uac desktop so you can't really read any message
why the banner then? :)
Anders
22nd November 2011 12:47 UTC
Originally posted by Yathosho
let's make this short (and go back to my initial post). that's exactly what i tried in the first place and that's also what lead to my initial post. the confusion starts with the correct irc server, the wiki mentions one, the forum another - both didn't work (one timed out, the other stated i had no authorization to connect)
Clicking the link @
http://nsis.sourceforge.net/Community and using the Firefox default handler works, I just tried it. (LoL and cockos are the same when it comes to IRC AFAIK)
Originally posted by Yathosho
why the banner then? :)
Like I said, it was to deal with loss of foreground focus.
MSG
22nd November 2011 13:02 UTC
I agree with Yatosho that since you're throwing a banner anyway, it'd be good to fill it with something that serves a purpose. "Executing setup with elevated rights..." seems like a good idea to me, it's good practice to let the user know what's happening.
As for IRC, yes, they're both the same server.
Yathosho
22nd November 2011 14:17 UTC
must've been a temporary issue with the irc server, i tried both: connecting manually and clicking the link to the channel. will try again when i get home, away from firewall land ;)
concerning the looks of the banner, i'd model it after the current unpacking dialog to avoid a "jumping icon"
Anders
28th November 2011 00:16 UTC
Added stupid invisible hack to the banner +other fixes and documentation in uac.nsh