Friday, January 9, 2009

And the survey says...

Well, we got quite a lot of user feedback in the survey that Casey ran on our site. I've read through a lot of it and thought I'd share some replies to specific comments. Keep in mind that we received many thousands of comments, so your exact comment may not be addressed here. With so many comments, it's very time-consuming to go through them all, so I just picked a few that stood out to me. At any rate, here we go:


"Could be clearer how to set up pidgin with ssh." - This request isn't quite clear, but I'm going to assume that the person wants to know how to use Pidgin either with an ssh tunnel or with ssh acting as a proxy. In the former case, ssh tunnels are documented quite extensively; you would configure your tunnel, then configure the Pidgin account to connect to localhost on the port you're forwarding. Using ssh as a SOCKS5 proxy is also well documented; you would set up the ssh proxy according to the documentation and then configure Pidgin's proxy settings to use the proxy on localhost. We haven't documented this ourselves because a lot of good generic documentation already exists, and it's not exactly a common use case that we see.

"I would like to see auto-reconnect be more resilient. I tend to have to manually reconnect after actual network outages. Also, it'd be nice if the encryption plugin knew when to get out of the way." - This is actually two comments, but I'll address them both. The auto-reconnect feature uses an exponential backoff that starts at just a few seconds between attempts and backs off to a maximum retry time of 34 minutes. If your Pidgin is compiled with NetworkManager support, NetworkManager may cause some delays due to not realizing the network is back. On Windows, there are some interactions with the Network Location Awareness features in Windows XP and Windows Vista that can cause delays in the reconnection as well. As for encryption, we didn't write the encryption plugins; that's a complaint/suggestion best taken up with whoever wrote the plugin you're using.

"Fix the chat input box please, or I'll switch to Carrier/Fun Pidgin." - Isn't Carrier dead? Anyway, if you feel the need to switch clients, that's your prerrogative. We've debated this at length on numerous occasions and we've decided to stick with the autoresizing input area, at least for now.

"Have the buddy list show up when i open pidgin, and have a popup when a person says something in a conversation." - There are two plugins already that provide popup notifications. One is called Guifications, the other is pidgin-libnotify. Windows users need to use Guifications. Users of other operating systems can use either. As far as the buddy list visibility goes, if you want to see the buddy list when you open Pidgin, either set "Show system tray icon" to "Never" in preferences, or have the buddy list open when you close Pidgin. The ability to start with the buddy list hidden was a feature request we chose to implement, and it would cause too many complaints for us to remove it.

"Better memory usage on windows or an up to date GTK included." - We do actively try to reduce Pidgin's memory usage where we can see room for improvement; patches are welcome for conditions we don't see. Our Windows person bundles new GTK+ runtimes as it's confirmed that new WinGTK+ releases don't introduce problems for us.

"crashes, crashes, crashes" - We do our best to fix the crashes that get reported. Perhaps you should read "Tips for Bug Reports" and then open a ticket.

"Docking pidgin in windows vista rearranges my icons on the desktop, but this is a problem with all programs that dock so not really important." - You've correctly realized on your own that Pidgin is not at fault here. This is a bug in Windows' desktop management, which we can't do anything about. I remember seeing this bug as far back as Windows 98, so it's not exactly new either. Maybe Microsoft could be convinced to fix this 11-year-old undesirable behavior?

"None. Keep it simple. I love that." - Thanks! We love simple. Clean, too!

"File transfer speed in MSN protocol" - We know a lot of our users are dissatisfied with the state of MSN file transfer support in libpurple. We're sorry you're not happy with it, but it's just not a priority for us. If someone comes along and implements a good (or even semi-reasonable) patch to do this, we'd be happy to work with that person to include better transfer support.

"Startup time. It takes about 10 seconds for me to get Pidgin 'usable'." - 10 seconds is not an unreasonable amount of time. Remember that Pidgin has to connect to all your accounts (MSN can be particularly slow here!), load your buddy list from the server, apply your saved preferences, load your saved plugins (including the protocol plugins), etc. There's a lot of work that goes on in those 10 seconds that is difficult, if not impossible, to streamline any further. You may, however, be able to shave some startup time off by removing any plugins you don't use (you can see paths and filenames by going to the Plugins window, clicking on a plugin's name, and clicking the expander by "Plugin Details").

"Some XMPP contacts are added twice." - We've had this bug forever, and it irritates us too. There has been some recent work on this. It should be better in 2.5.3 and 2.5.4 than in previous releases.

"1) One thing that confuses me...when a pidgin window aside from the chat window has focus (Buddy List, Manage Accounts, etc), and you type, it opens up a little window with that text in it...I probably just need to RTFM and it's working as intended, but who knows... 2) When setting status to Away, you are allowed to enter a message. Prior to version 2.5.2, when setting status back to Available, this message would be cleared automatically. From version 2.5.2 and after, it is now highlighted for manual deletion, which I find annoying. 3) Animated smileys would be awesome (but I assume that is more so on Hylke). I do love the current smileys though!" - Three points, so three answers: 1. This is a feature provided by GTK+ sometimes called "typeahead search." If you type text that matches a buddy's name or alias, that buddy will automatically be highlighted. 2. We discussed this a couple of times. Both use cases are valid, but it's our opinion that it's less annoying for people who want new messages to have to type a single backspace than it is for users who want to keep the same status message across multiple statuses to have to retype their whole message. Sorry. 3. Yes, animated emoticons would be something Hylke would need to create for us if he felt so inclined, but I think we'd all be better served if animated emoticons were a separate theme from the default.

"GTK" - I'm going to go out on a limb and say the person who wrote this complaint is a Windows user. We use GTK+ because Windows was not our first platform, and GTK+ was the best toolkit for the purpose when Pidgin was started in 1998. We would love to see someone write a native Windows interface for libpurple, as we know it would serve most Windows users better than our GTK+ interface ever could.

"It is not very easy to install Pidgin on Linux distributions, especially because of dependencies." - This is true of source-based installation of Pidgin for those who don't compile many applications themselves or those new to the world of compiling applications. Pidgin is provided in a pre-packaged, dependency-tracked format by all the major Linux distributions, as well as through the BSD ports system, Fink on Mac OS X, and Macports on Mac OS X. All of these distributors provide an extremely simple way to install Pidgin.

"Using special characters in statuses crashes Pidgin. With MusicTracker, this can happen a lot... and crash my client. Pidgin ought to replace these characters with question marks or something. Also, file transfer is very slow and erratic, especially with non-Pidgin users." - I've covered the file transfer above, to an extent. Special characters themselves don't crash Pidgin. MusicTracker could probably prevent some crashes by making sure that all strings it passes into libpurple are UTF-8. We have recently made some changes that should help in this regard, but plugins should still make sure they hand us only valid UTF-8 strings.

"Pidgin steals keyboard focus when it raises a conversation window, which is very bad. I want windows to raise when I receive new messages, but without stealing keyboard focus, especially from another conversation window I'm actively typing in...." - I have trimmed a longer user submission here to a specific issue to reply to for clarity and brevity. Pidgin doesn't ask for focus. Some window managers, however, treat a raise request as a focus request. These window managers should be fixed. Trying other window managers and seeing their behavior would confirm this.

"I'd like to see some new looks available by themes (sounds included would be nice) so long as they don't change the position and location of various options. I know skins and themes are mentioned already in the survey but I've wanted them for a while and it's the only thing I can think of that I really want to see." - Some theming support, including sound themes, was worked on during the Google Summer of Code for 2008. We expect to release this work in Pidgin 2.6.0, but no promises.

"Wrong encoding for incoming messages from 'QIP' users (icq protocol) writing with diacritic marks in czech/slovak (encoding central european, both iso 8859-2 and win 1250)." - I really wish everyone would just use UTF-8 already. All these problems would magically go away.

"x-status support for ICQ protocol" - This is something else we expect to release in Pidgin 2.6.0.

"doesn't always connect to a network" - More details are necessary to help in this case.

"A two pane plugin-preferences dialog would be nice. You select the plugin on the left pane and the preference are displayed in the right pane." - It seems that no matter what we do for plugin configuration, no one's ever happy with it. If someone can come up with a plugin configuration dialog that everyone can live with, I'd gladly fight (if needed) to get it in a release.

"Please fix the auto-away option - it's been broken since at least version 2.1 or 2.2. For example, when status is set to 'Invisible', it changes to 'Away' after been idle for a while (this shouldn't happen when offline) - and then, when initiating a conversation with someone, it changes back to the original offline status ('Invisible' in this case)! Peer-to-peer MSN file transfers would also be nice, as that function is almost unusable in its current state." - Again, I have already covered MSN file transfers, but I'm responding to this for the auto-away complaint. I'll admit that the behavior you describe is undesirable, but it is working as designed. It might make sense to exempt an original status of "Invisible" from auto-away, but this is a slippery slope in some respects--we make this change for Invisible, and people will want all sorts of unique exceptions.

"Too many dependencies, make it lighter!" - Yes, we have a lot of dependencies, but very few of them are requirements. If you're building Pidgin yourself, you can disable all but the absolute minimum required dependencies by making use of the "--disable-????" family of arguments to the configure script ("./configure --help" will show you the options). We are very cautious about adding new dependencies unnecessarily.

"MSN protocol seems buggy, IRC isn't very customizable" - The MSN protocol plugin has had some major cleanup work done on it recently. In Pidgin 2.5.4, MSN support should be better than it was in 2.5.2 without the crashes we had in 2.5.3. The IRC protocol is somewhat unique in that it doesn't really fit well into Pidgin's IM-centric model. Pidgin really isn't designed to be a fully-featured IRC client; however, recent releases of Pidgin have functionality which allow plugins to interact with the raw IRC protocol data; someone so inclined could write plugins to extend IRC functionality.

"Chat timestamps on same line as name (not have a hole line devoted to timestamps." - Make your conversation window bigger or turn off some plugins to find the culprit, as there are a few that might be capable of doing that.

"I still miss protocol icons. Green "balls" suck hard." - Sorry, but this is something we're always going to disagree on.

"MSN extra forming like this: [c=61]NICK[/c][c=55]" - We currently have no intention of supporting this "feature" of MSN Plus. I would certainly be open to a patch that just stripped that garbage out, though.

Feature Requests

"An 'official' portable version." - We already provide one. Configuring a portable installation is explained in the FAQ.

"Close all the bugs. That's a major feature." - Oh, how I wish that were true.

"Skype support" - It's our opinion that Skype integration into Pidgin is not possible in a legal manner.

"Being able to add your own emoticons in msn" - This was added in Pidgin 2.5.0. See the Tools menu, and look for "Smiley."

"link to open mail inbox, even when there are no new emails to read." - At least our Yahoo! plugin provides this feature (Accounts->Your Account->Open Inbox). MSN does too, via Accounts->Your Account->Open Hotmail Inbox.

"Video and Voice" - This is currently in the works. We hope to have this out for 3.0.0 (no, we don't know when that will be).

"Major Changes to the interface should be supported as plug-ins. Functionality should NEVER be removed from a product unless that same functionality is re-included via a plug-in or an interface option." - This is, of course, your opinion. Our opinion is that if we were to do this, we'd have an unmaintainable mess of options and plugins. It would be impractical to implement and impossible to manage.

"I'd love to be able to set different notification styles for different groups/users; something like setting different ring tones on cell phones. Also along these lines, I'd love to be able to show different groups different status messages; for instance, if I'm at work and appear offline to everyone but my work buddies group." - Notification styles on a per-group or per-buddy basis is possible by using the Guifications plugin and installing and loading multiple themes. Then you can right-click a group or buddy and you'll find an option to change the Guifications theme for that group/buddy. The status messages per-group could be interesting if all the protocols supported it. However, to the best of my knowledge, no protocol (except possibly XMPP) supports such a capability.

"A Native Windows Interface or QT4 interface (also ports over to windows / Linux / MacOS quite easily from what I understand)" - We have no intention of developing such interfaces ourselves, but we would love if someone made a libpurple interface that fit natively into a Windows desktop environment. We certainly have no objections to someone creating a libpurple interface with Qt, either. None of our current development team has any interest in starting these projects, but I'm sure we're certainly willing to help (to the best of our abilities) with any issues we can.

"Expanded SIP support to replace Windows Messenger 5.1 (not Live Messenger!) since I use it all-day at work." - None of us use Microsoft's corporate instant messaging software, so it's difficult at best for us to make sure that such a protocol plugin is maintained. We also really don't want to pick up any additional protocol support, given our history of protocol maintainers disappearing. There is a team of developers working on the SIPe plugin with the explicit purpose of working with Microsoft's products. Perhaps you should investigate this plugin.

"Better password security. Please, encrypt the passwords inside your config files." - We have an entire wiki page devoted to this topic. Encrypting passwords in accounts.xml is pointless, as it would be trivial to use our own code in an attack. Further, password security is one of the topics of our most recent participation in the Google Summer of Code. We're hoping to have the work that was done on it released soon.

"more support for 3th party protocols like xfire etz meaning box release with these protocols etz" - Um, no. I already discussed the reasons we don't accept additional protocol support.

"Protocols as PlugIns (maybe selectable witch should be installed), AutoUpdate (not only tell about new Versions, also [if user agrees] download, close the client and open setup)" - Protocols have been plugins for many years now. We don't really see a point in making it possible to choose which protocols to install with the Windows install program. Automatic updates would only be useful on Windows, and could actually be done with a plugin if someone were so inclined to write one. We don't have any interest in making it happen though. Sorry.

"It'd be nice to see a version for Windows Mobile... maybe this is already out there and I need to Google... ?" - Pidgin requires GTK+ and Glib (among other libraries), neither of which are available for Windows Mobile yet. Pidgin and libpurple are also a bit heavy in terms of memory and processor time consumption for most devices that run Windows Mobile. You'd be muchbetter served by a client designed specifically to work within the limitations of the Windows Mobile platform.

"Skype support (I've tried the plugin, but it requires Skype client to be running and has no video support, so it's worthless)" - I've covered Skype support already, but I'm going to again point out that we feel this plugin is a violation of our license terms and should not be used by anyone.

"Facebook chat support would be an awesome thing. I'm using the plugin now, but it's got some quirks." - There was a discussion on the mailing list about Facebook support. We came to the conclusion that we don't want to deal with it. If and when Facebook ever rolls out their XMPP interface, we'll be able to support Facebook Chat effortlessly.

"Organize contact list by user by protocol. This way I have a single entry for the user and expand them for multiple ways to contact. It would clean up my contact list significantly." - We have supported combining buddies into contacts for years. Perhaps you should take a look at this FAQ entry?

"Integrated music player support for changing the status based on what I'm listening to...." - I trimmed this down too to get to the important point here. We don't want this in Pidgin. People wanting stuff like this is exactly why we have a plugin system.

"Add a spell check feature!" - We've had one for years. Install an appropriate dictionary for the language you're running Pidgin in.

"1. Support for buzzing/nudging 2. Mass inviting to chats 3. ability to receive MSN voice clips" - 1. We've had this support for years. Use the /nudge command on MSN and /buzz on Yahoo. 2. As soon as someone comes up with a sane UI for it, we'd be happy to accept the patch implementing it. 3. I personally wouldn't mind if this never landed in libpurple, but again this is something that would have to be implemented by someone who wants it.

"Being able to update Pidgin without having to reinstall the whole thing." - That's precisely what an update is--the whole application is replaced with a new version. What about this is unacceptable?

"Multiple user support. Set up a master password for each user to access their accounts." - We've discussed this on numerous occasions. We are not interested in implementing something that the operating system already handles perfectly with its native user account support.

"A plugin for general IMAP mail check. frameless contact list window like adium offers on OSX." - Pidgin is an IM client, not a mail client. That said, anyone who wants this feature is welcome to write a plugin to implement it. Frameless buddy list windows are much easier said than done. Remember that Adium supports precisely one platform, Mac OS X, and as such can do whatever they want to fit better into their environment. Given that we support more platforms than Adium, it makes it particularly difficult because we'd have to have hacks in the code for every platform we support, as well as multiple variations of each plugin. There's not enough benefit to warrant the increase in complexity and the inclusion of such ugly hacks.

"Database logger. There was a SoC project out but it never got completed. Probably due of developers not accepting the student's patches? Support for mIRC srcipts." - If you're referring to the "remotelogger" SoC project, there are some adjustments necessary, and ideally we'd be able to kill off some of the API that it supersedes, thus requiring a 3.0.0. It is, however, in our monotone repository. As far as mIRC scripts, again, this could be done by a plugin. We have no intention of having an internal scripting engine just for mIRC scripts.

"* Cook breakfast" - If Pidgin could cook my breakfast for me, I'd save a LOT of money by not needing to go somewhere for breakfast when I'm too lazy to cook for myself.

"Encryption for chatlogs, auto updating." - This can be provided better by the operating system than by anything we could possibly implement. NTFS filesystems on Windows support encryption, and there are many encryption options for Linux and UNIX environments. Storing your .purple folder in an encrypted area (by symbolic link/junction point or by wrapping Pidgin with a script to point to a non-default configuration directory) would thus give you encryption on your logs. I've already covered automatic updates.

"My opinions refers to Pidgin 2.5.2. Spanish support please and other languages." - We have an active Spanish translator. We have a total of 73 translations, over 2/3 of which are very actively maintained.

"A plugin manager that you could install plugins with, without having to manually place the dills inside of the folder would be nice. Something simialr to mozillas addon managers." - I actually have a plan for something like this, but I haven't been motivated to implement anything for it yet.

"UPDATE OPTION! Once I was running like 2.3.1, when 2.5.2 was out, and I had no idea, pidgin does not have any update monitor to let me know when updates are out, the actual process is simple and i love it." - Take a look at the Release Notification plugin that's included with Pidgin by default.

"More accessible extensibility. Having to write plugins in C and compile them is out of reach for many folks who would write decent plugins." - Perl and TCL plugins are already possible. On Linux/UNIX systems, it's possible to interact with libpurple over D-Bus as well.

General Feedback

"Please switch from Monotone to a wider-known DVCS like Git, Bazaar or Mercurial." - We chose our version control system very carefully several years ago. We went with what we felt was the best choice at the time, and currently, the effort involved in a switch incurs a much higher cost than it's worth.

Why the name change? (I paraphrased this question from several statements) - We entered into an agreement with AOL to avoid being sued for infringement on the "AIM" trademark. The agreement required a name change. This was discussed at length two years ago when we made the change. For the record, we can no longer use our former name, even in technical discussions. This too was part of the agreement.

"I've been always been happy with this product and think it is a fine example of Open Source software and Free Software ideals. Thank You." - And thank you for the kind words. There were a lot of people echoing this sentiment in the general feedback section. In fact, there were far more of this type of comment than I expected.

"Try not to be jerks when people are providing feedbak or suggestions trough mailing list." - It's always interesting to see this come up, because most of the time when we get these complaints, I find that our reactions are quite justified. I'd also like to point out that if you don't like us, you don't have to interact with us. I'm sorry, but we don't exist just to babysit all our users. We have over three million users, and we just can't do it for all of them, or even for a fraction of them.

"The development process should be more transparent, we the users have no idea what's coming next (and when), and if we want to find out, we have to browse dozens of pages of bug reports or some weird wiki page. I'm thinking about the whole 'MSNP15/Personal Message' thing, especially." - We don't know what's coming next or when it will show up either. The development process is transparent; come to our IRC channel, our XMPP chat, and the development mailing list. We have a lot of very public discussion on what we should and shouldn't do and what direction development should go in.

"Don't get me wrong, I appreciate all the effort made by Pidgin. But it seems to me the strong feelings of a few are getting in the way of a real, quality IM client. I understand the desire to remain lean and modular, but that only works up to a point. Of course voice and video chat make more sense as being plugins, but must there really be a plugin just to enable "hiding the buddy list when it's created" ? Pidgin is too complex for the average user. Firefox has struck a moderately good (I'll never say perfect) balance of what basic features are required for any browser and which ones require extensions. I think Pidgin can do the same for IM clients. I'm always looking forward to new releases. Thank you for your work." - Complexity is relative, and in the case of a web browser, it is significantly clearer what belongs in the application itself and what requires extensions/plugins. In an IM client, it's not so clear. This is why we have a plugins system--things we think don't belong in Pidgin itself can be developed as plugins. The strong opinions I and others express are a check and balance to make sure we don't as a project go to extremes, and also serve to keep us focused on important things like fixing bugs instead of pointless political stuff and catering to every person's whim. The thing a lot of our users forget quite easily is that we work on Pidgin for ourselves, so if we don't like an idea, we're going to rally against it and block its inclusion.

There are many more comments I'd have liked to address here, but there are just too many of them and I don't have the time to draft individual responses to all of them. I hope these responses show that we do listen to user feedback. We always listen, but we don't always act. There is a difference there that a lot of people don't understand.

The sheer number of responses we received, including the comments expressing what a good job we've done, have shown me that we were correct when we asserted that we had a huge portion of our userbase that is either indifferent to many of the changes we make or mostly satisfied with us in general and don't feel the need to send feedback.

In short, thanks for the feedback!