You can subscribe to this list here.
| 2011 |
Jan
|
Feb
|
Mar
|
Apr
(5) |
May
(1) |
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|---|
|
From: Johnny D. <kc...@ju...> - 2011-05-03 06:42:22
|
Sending out this email because I have yet to receive a single message to the list. I'm not sure why, so I'm seeing if sending one from my spam address, which is not registered with the group, will cause me to receive it. Johnny |
|
From: Me <kc...@gm...> - 2011-04-13 18:56:11
|
I definitely agree that some sort of automatic converter system would be excellent to have. Forwarding to the list. Johnny On Wed, Apr 13, 2011 at 11:42 AM, Shane Kamar <sha...@ma...> wrote: > I see your point. As for the plugin system, the biggest hurdle, to me, is > merely the fact that not all users will have identical configurations and > thus I wish to submit that perhaps some flexibility or automation could make > the user experience more streamlined. And also, you're right the conversion > process is complex and beyond the scope of this project. > > I am working on a mac and thus have XCode installed alongside the Qt > extensions and I know I have CMake on here as well. Boost is also on my > system, although I cannot remember at the moment which folder it is stored > in. > > I apologize for bringing up "issues" outside of the scope of the project. > I am familiarizing myself with the source code. I do hope you will consider > the idea about the automatic downloading of prerequisites (i.e., ffmpeg, > etc.) as part of the polishing process, as while much of the user base may > be quite skilled with system maintenance, there still exists a large base > largely unfamiliar with acquiring and compiling prerequisites, especially > without instructions. > > Shane > > On Apr 10, 2011, at 2:56 PM, Johnny Dickinson wrote: > > Do you have our existing code compiling yet? We already have a fully > functional plugin system. The speed cost due to being object oriented really > isn't a concern, since we don't perform the actual conversion ourselves. The > bottleneck is the converter process, which we cannot control. > In order to build the source code, you will need boost, qt development > libraries, and cmake. You should familiarize yourself with our existing > source code, a lot of your concerns really aren't problems. Our program is > already a fast, functional converter. At our current stage of development it > is more about polishing it to make it really usable. > > Johnny > > On Apr 10, 2011, at 3:33 PM, Shane Kamar <sha...@ma...> wrote: > > Johnny, > > Thank you for your response again. It's understandable that people become > busy. E-Mail is a fine form of communication. Do any of you also have Skype > or some type of IM application? > > I noticed the Qt GUI. Personally, what I would recommend is having > individual (separate but related sub-projects) for each respective *major* > OS system--that is, a native Mac (Cocoa) interface, a .NET WPF Windows > Application, and either a KDE/GNOME/GTK or wxWidgets for all other platforms > or just maintaining Qt for otherwise miscellaneous systems. > > As for the debugging, I have been delving into the header files on my Mac > system and found them to be largely POSIX compliant, and believe that for > *nix/BSD systems we might could use an in-house debugging system with macros > in cpp to run "rusage", "ptrace", etc. and their Windows equivalents if > debug is set. This could then be redirected to a logging system. > > I noticed the mixing of Qt libraries with the core system. In a moment I > will discuss further my idea with the plugin system and address your > concern. At the moment, I wish to venture forth that the core be perhaps > written in (non-objective) ANSI C code. > > I have been reading up on several experts opinions on object-oriented > programming paradigms and find that for such a processor-intensive function > such as video/batch conversion, it would be more efficient and faster to > implement a imperative paradigm in a language such as C: that is, splitting > data structures from their respective methods in order to avoid excessive > encapsulation and memory. > > As processor speed increases, unfortunately memory access speeds have only > increased in a small amount versus processor speeds. Furthermore, > audio/video processing can be even more quickly facilitated by accessing > GGPU functions, passing off these memory-intensive operations to the GPU > systems (libraries exist for multiple GPUS, the most popular being CUDA from > nVidia). Thus, with C we can better leverage system resources for speed, > efficiency, and minimum intrusion on other processes. > > I understand your concerns with LAME binaries. Not all of them are public > domain, but many can be obtained freely by the end user. Having said that, > the plugin system could easily facilitate this, in the following > for-instance. > > Each plugin bundle (probably best to be tar/gz) will be structured to > include necessary file structures, as demonstrated here in the attached > image. There will be a meta file: a binary file structure read and wrote to > from C structures for efficiency. There will be a utility for developers to > "compile" their XML files into binary meta files. > > There will be a folder which will contain scripts for ensuring all > prerequisites are either there--that is, downloading LAME binaries on behalf > of the user to a set location) and also a folder containing binaries for > each platform (scripts to check for preexistence or the binaries themselves, > as well as "dll" or "dylib"'s for set platforms). By offering the > possibility to use scripts, we give the option of either packaging small > libraries or locating existing libraries or downloading them off of the > internet. > > Also, you seem concerned about avoiding writing code for new converters. > This brings up a very interesting point and a chance to talk about expanding > the project. Although XML manifests may seem a viable solution, consider > that not every person shall have their system set up in such a way where > required binaries are in the same location--scripts can perfectly locate or > acquire prerequisites. Such platform-dependent binaries would be small and > faster to read, load, and use than an XML system that would require the user > to have everything installed in one spot. > > By enhancing the plugin system, more flexibility can be achieved: data can > be processed at a higher rate by using intelligent multi-threading as well > as implementing CUDA for lightning fast conversions. Some plugins would read > certain filetypes, and some would contain the specification for rewriting > files. This would be more intuitive than calling arbitrary binaries without > video acceleration, etc. > > I do indeed have a development environment set up: on my Mac I use X-Code > 4, which is a very able multi-language (mainly c-based) IDE supporting C and > Qt. On the Bootcamp partition I have Microsoft Visual Studio 2010. > > I see high potential for this project. I am very excited about working on > it with you. At the moment, I am drafting a proof-of-concept debugging > system and plugin system in C in X-Code. Also, I am looking into video > conversion algorithms. > > What are your further thoughts now? I am curious to know how you feel about > these ideas. > > Best, > > Shane > > On Apr 10, 2011, at 1:49 PM, Me wrote: > > Shane, > > It looks like you have a lot of great ideas. Sorry I didn't respond > earlier, I had a bit of a crazy end to the week. I want to include the other > developers in the discussion, so I CC'd the mailing list. I just set it up, > but I think email might be a good way to communicate. Does that work well > for you? I think I prefer it to the forum for development discussion. > Sourceforge's forum isn't exactly stellar anyway. > > As far as the GUI goes, currently we just have the Qt GUI. It was chosen > because it is pretty functional on Windows/Linux and works with OSX, unlike > GTK. Delving into the code a bit, currently we have some Qt code in > BackendManager. This code primarily serves to send debugging information to > the GUI. I want to refactor all of that code out into it's own class. I was > thinking something along the lines of a generic logger class, with our > current Qt code inheriting that as an implementation. Once that code has > been pulled out, we should be able to build the backend code as a standalone > library that only depends on Boost. I think it is important to have the > backend separate so that it can be used as a converter engine for any > program. I was waiting to implement that because Alex is currently > implemented multithreaded loading in BackendManager, which involves some > significant changes to it's operation. Once the backend code is no longer > explicitly tied to the frontend, we will be able to do things like write > alternate UI's. I have done some work on a command-line interface, which I'm > using to provide a test framework for cleaning up the backend Qt dependency. > > As far as the plugin system goes, one concern we had was patent issues. I > don't think we can legally distribute LAME binaries without owing licensing > fees. However, some sort of system like you suggest would be excellent for > providing drop-in format support, particularly on Window and Mac. My goal > with the XML converter description was to avoid having to write any code to > support a new converter, so hopefully we won't have to worry about providing > a cross platform library in any package. > > Your suggestion of a server-based converter is interesting, I hadn't > thought of that. However, I think we need to work on core functionality > first, as there are several areas where we are still lacking as a converter > program. Once that is done, it would be very cool to be able to support > "cloud converting" or similar. > > We already have a cross-platform build system, CMake. I don't know if it > supports cross-compiling. Setting up a build server to automatically build > releases should be doable, although I'm not sure that our code has reached a > maturity level to make that worthwhile yet. > > I'm glad you have so many great ideas. Do you have a development > environment set up yet? Let me know if there are any issues I or the other > developers can help you with. Think about picking one improvement to work > on, and starting from there. If there is anything that isn't clear in the > code, don't hesitate to ask. > > Thanks, > Johnny Dickinson > > On Fri, Apr 8, 2011 at 1:59 PM, Shane Kamar < <sha...@ma...> > sha...@ma...> wrote: > >> Thank you for your timely reply! >> >> I am indeed a Mac user; however, by the miracle of bootcamp and parallels, >> I develop also for Windows (7) as well as POSIX (suseLinux). >> >> I personally love Mac--as well as the user interface conventions that >> Apple has pioneered--and would love to further streamline UBC inside and out >> to help it run smoothly with an interface equally beautiful across >> platforms; that being said, for the best integration, a few forks on the >> G/UI level may be necessary, as each platform offers an abundance of system >> integration methods not exposed in QT. >> >> For example, Windows Vista introduced Windows Presentation Foundation, >> which streamlines the creation of UI's in the style of the native (Windows) >> operating system; with Windows 7 came several new elements not available >> across platforms, such as Taskbar options, progress displays, etc. >> >> Equally, KDE and GNOME carry their own user interface elements, some of >> which are exclusive their respective environments. >> >> Naturally, Mac OS X has been a study in user-interface ingenuity in the >> eyes of developers and design critics alike. The latest iterations of the >> popular Apple OS have brought stylish components such as Quartz--a graphical >> programming language for laying out fast, runtime-rendered, smooth graphical >> interfaces--as well as Core Graphics--a similar framework for >> Objective-C--and various integrations with the operating system. >> >> With the growing movement of processing power from desktop (laptop) >> computers to scalable server environments, I would just like to venture >> forth the idea of having a host-able (server-run/headless) distribution of >> UBC alongside the current desktop version. >> >> I have taken a precursory glance at the distribution as well as the code. >> Being a mac/posix/windows user, I think it would be advantageous to set up >> an automated cross-compiling build system, or failing that, an automated >> distribution system that would compile appropriate redistributable s for >> each appropriate platform. >> >> Also, I realize I am new to this, so I hope my ideas aren't taken as >> arrogance, but I just though I would mention a thought I had on plugins. I >> like your XML plugin system, and I think it would make it ever more flexible >> if we were to implement a kind of "bundle" or "container" system for >> plugins; Mac OS X marks certain folders as "bundles" and they are thusly >> treated as a single file. Perhaps we could do something similar in the way >> that Apple handles documents and iOS applications. >> >> An iOS application is just a folder containing everything that the app >> needs to run, all zipped up with a manifest. >> >> Perhaps for the plugin system, we could do something similar: each plugin >> could be a ZIP file containing: 1) executables for the appropriate tool(s) >> needed for that plugin or group thereof; 2) the manifest (XML, but with type >> declarations; 3) any (multi-platform) DLL or .so or .dylib that could be >> used to plug-in the the main program. This would allow great flexibility >> across platforms while assuming minimal requirements on the client-side. >> >> I hope these thoughts prove useful, and I very much look forward to >> working with you on this wonderful project. Please let me know what your >> ideas on my suggestions and observations. >> >> Unfortunately it's time for Linear Algebra so I must send this email now >> and rush to class. >> >> I will further peruse the source code after my 2PM class (it's 1PM here), >> so I will be sending another email shortly thereafter. >> >> Thanks again, and best wishes, >> >> Shane Kamar >> >> On Apr 8, 2011, at 11:36 AM, Me wrote: >> >> Shane, >> >> We would love to have you. In particular, based on your email I'd venture >> to guess that you are a mac user. We have been trying to work with mac in >> mind, but none of the developers have a system running OSX, so we've had to >> rely on information from a couple people who tried running it once. If you >> wouldn't mind ensuring we have a good experience running on OSX, that would >> be wonderful. >> >> >> Just a quick overview of the project so far: >> The other active developers and myself are undergraduate students at >> Rensselaer Polytechnic Institute. >> We are developing with the Rensselaer Center for Open Source >> <http://rcos.rpi.edu/>http://rcos.rpi.edu. >> >> Git usage: >> As far as development goes, we are still in fairly early stages of >> development. >> We develop new features in separate branches. These can be pushed to the >> sourceforge server for collaboration. >> We have two main branches, devel and master. >> Our policy is that the devel branch should compile, without major breakage >> if possible. >> The master branch should be usable, so that building it produces a working >> program. >> When we have a stable program with new features, we want to release. >> So far we've been pretty flexible about sticking to these requirements, >> both because we've been trying to get our program to the stage of being >> ready for general use, and producing demo code to get our program out in >> the wild. >> >> Previously, we have done a lot of planning offline, as we have face to >> face interaction on a regular >> basis. We will make an effort to get our discussion online in order to >> keep you in the loop. >> We have a small amount of documentation on our wiki and bug tracker, but >> it is incomplete. >> I also built doxygen documentation from the source code, although it is >> out of date because currently >> I build and upload it manually. We also have a simple task list on our >> website, which makes it easy >> to list what we're currently working on. Again, we haven't been as good >> about keeping them up to date >> as we should be because we meet face to face, but we will push to get our >> work online. >> >> To get started setting up a development environment, visit: >> >> <http://sourceforge.net/apps/mediawiki/unibatchconv/index.php?title=Developer_Documentation> >> http://sourceforge.net/apps/mediawiki/unibatchconv/index.php?title=Developer_Documentation >> >> Post questions you have about the code on the forum (which we can start >> using finally): >> <http://sourceforge.net/projects/unibatchconv/forums/forum/1231106> >> http://sourceforge.net/projects/unibatchconv/forums/forum/1231106 >> >> Just so you are aware of our contributions: >> I'm einarin, I developed Converter and the XML converter specification. >> Alex Hunt (FudgeMutator) developed the conversion management code and most >> of BackendManager. >> Frank Kotarski (Kotarf) developed the multistage conversion code. >> Jon Kriss (Jonnyk127) developed the GUI. >> >> Feel free to email me with any questions. I look forward to working with >> you. >> >> John Dickinson (einarin) >> >> >> Shane Kamar >> Sophomore, Auburn University Honors College >> CSSE, College of Engineering >> < <sc...@au...>sc...@au...> >> (256) 682-2032 (mobile) >> >> >> > > Shane Kamar > Sophomore, Auburn University Honors College > CSSE, College of Engineering > < <sc...@au...>sc...@au...> > (256) 682-2032 (mobile) > > > > Shane Kamar > Sophomore, Auburn University Honors College > CSSE, College of Engineering > <sc...@au...> > (256) 682-2032 (mobile) > > > |
|
From: Me <kc...@gm...> - 2011-04-12 19:14:13
|
If you are interested in helping, we can use more developers. If you're interested in helping, pull our code and take a look around. We have a wiki page on setting up a development environment, if you need help or have questions feel free to email me. Take a look at our bug tracker for open issues for an idea of things to work on. Feel free to email me or the mailing list. Thanks, Johnny Dickinson On Tue, Apr 12, 2011 at 12:11 AM, <mor...@us...> wrote: > Hey I know some C++ but I'm looking for experience I'd > definitely be up to help out in any way I can. > > -- > This message has been sent to you, a registered SourceForge.net user, > by another site user, through the SourceForge.net site. This message > has been delivered to your SourceForge.net mail alias. You may reply > to this message using the "Reply" feature of your email client, or > using the messaging facility of SourceForge.net at: > https://sourceforge.net/sendmessage.php?touser=2053105 > > |
|
From: Johnny D. <kc...@gm...> - 2011-04-10 19:55:50
|
Do you have our existing code compiling yet? We already have a fully functional plugin system. The speed cost due to being object oriented really isn't a concern, since we don't perform the actual conversion ourselves. The bottleneck is the converter process, which we cannot control. In order to build the source code, you will need boost, qt development libraries, and cmake. You should familiarize yourself with our existing source code, a lot of your concerns really aren't problems. Our program is already a fast, functional converter. At our current stage of development it is more about polishing it to make it really usable. Johnny On Apr 10, 2011, at 3:33 PM, Shane Kamar <sha...@ma...> wrote: Johnny, Thank you for your response again. It's understandable that people become busy. E-Mail is a fine form of communication. Do any of you also have Skype or some type of IM application? I noticed the Qt GUI. Personally, what I would recommend is having individual (separate but related sub-projects) for each respective *major* OS system--that is, a native Mac (Cocoa) interface, a .NET WPF Windows Application, and either a KDE/GNOME/GTK or wxWidgets for all other platforms or just maintaining Qt for otherwise miscellaneous systems. As for the debugging, I have been delving into the header files on my Mac system and found them to be largely POSIX compliant, and believe that for *nix/BSD systems we might could use an in-house debugging system with macros in cpp to run "rusage", "ptrace", etc. and their Windows equivalents if debug is set. This could then be redirected to a logging system. I noticed the mixing of Qt libraries with the core system. In a moment I will discuss further my idea with the plugin system and address your concern. At the moment, I wish to venture forth that the core be perhaps written in (non-objective) ANSI C code. I have been reading up on several experts opinions on object-oriented programming paradigms and find that for such a processor-intensive function such as video/batch conversion, it would be more efficient and faster to implement a imperative paradigm in a language such as C: that is, splitting data structures from their respective methods in order to avoid excessive encapsulation and memory. As processor speed increases, unfortunately memory access speeds have only increased in a small amount versus processor speeds. Furthermore, audio/video processing can be even more quickly facilitated by accessing GGPU functions, passing off these memory-intensive operations to the GPU systems (libraries exist for multiple GPUS, the most popular being CUDA from nVidia). Thus, with C we can better leverage system resources for speed, efficiency, and minimum intrusion on other processes. I understand your concerns with LAME binaries. Not all of them are public domain, but many can be obtained freely by the end user. Having said that, the plugin system could easily facilitate this, in the following for-instance. Each plugin bundle (probably best to be tar/gz) will be structured to include necessary file structures, as demonstrated here in the attached image. There will be a meta file: a binary file structure read and wrote to from C structures for efficiency. There will be a utility for developers to "compile" their XML files into binary meta files. There will be a folder which will contain scripts for ensuring all prerequisites are either there--that is, downloading LAME binaries on behalf of the user to a set location) and also a folder containing binaries for each platform (scripts to check for preexistence or the binaries themselves, as well as "dll" or "dylib"'s for set platforms). By offering the possibility to use scripts, we give the option of either packaging small libraries or locating existing libraries or downloading them off of the internet. Also, you seem concerned about avoiding writing code for new converters. This brings up a very interesting point and a chance to talk about expanding the project. Although XML manifests may seem a viable solution, consider that not every person shall have their system set up in such a way where required binaries are in the same location--scripts can perfectly locate or acquire prerequisites. Such platform-dependent binaries would be small and faster to read, load, and use than an XML system that would require the user to have everything installed in one spot. By enhancing the plugin system, more flexibility can be achieved: data can be processed at a higher rate by using intelligent multi-threading as well as implementing CUDA for lightning fast conversions. Some plugins would read certain filetypes, and some would contain the specification for rewriting files. This would be more intuitive than calling arbitrary binaries without video acceleration, etc. I do indeed have a development environment set up: on my Mac I use X-Code 4, which is a very able multi-language (mainly c-based) IDE supporting C and Qt. On the Bootcamp partition I have Microsoft Visual Studio 2010. I see high potential for this project. I am very excited about working on it with you. At the moment, I am drafting a proof-of-concept debugging system and plugin system in C in X-Code. Also, I am looking into video conversion algorithms. What are your further thoughts now? I am curious to know how you feel about these ideas. Best, Shane On Apr 10, 2011, at 1:49 PM, Me wrote: Shane, It looks like you have a lot of great ideas. Sorry I didn't respond earlier, I had a bit of a crazy end to the week. I want to include the other developers in the discussion, so I CC'd the mailing list. I just set it up, but I think email might be a good way to communicate. Does that work well for you? I think I prefer it to the forum for development discussion. Sourceforge's forum isn't exactly stellar anyway. As far as the GUI goes, currently we just have the Qt GUI. It was chosen because it is pretty functional on Windows/Linux and works with OSX, unlike GTK. Delving into the code a bit, currently we have some Qt code in BackendManager. This code primarily serves to send debugging information to the GUI. I want to refactor all of that code out into it's own class. I was thinking something along the lines of a generic logger class, with our current Qt code inheriting that as an implementation. Once that code has been pulled out, we should be able to build the backend code as a standalone library that only depends on Boost. I think it is important to have the backend separate so that it can be used as a converter engine for any program. I was waiting to implement that because Alex is currently implemented multithreaded loading in BackendManager, which involves some significant changes to it's operation. Once the backend code is no longer explicitly tied to the frontend, we will be able to do things like write alternate UI's. I have done some work on a command-line interface, which I'm using to provide a test framework for cleaning up the backend Qt dependency. As far as the plugin system goes, one concern we had was patent issues. I don't think we can legally distribute LAME binaries without owing licensing fees. However, some sort of system like you suggest would be excellent for providing drop-in format support, particularly on Window and Mac. My goal with the XML converter description was to avoid having to write any code to support a new converter, so hopefully we won't have to worry about providing a cross platform library in any package. Your suggestion of a server-based converter is interesting, I hadn't thought of that. However, I think we need to work on core functionality first, as there are several areas where we are still lacking as a converter program. Once that is done, it would be very cool to be able to support "cloud converting" or similar. We already have a cross-platform build system, CMake. I don't know if it supports cross-compiling. Setting up a build server to automatically build releases should be doable, although I'm not sure that our code has reached a maturity level to make that worthwhile yet. I'm glad you have so many great ideas. Do you have a development environment set up yet? Let me know if there are any issues I or the other developers can help you with. Think about picking one improvement to work on, and starting from there. If there is anything that isn't clear in the code, don't hesitate to ask. Thanks, Johnny Dickinson On Fri, Apr 8, 2011 at 1:59 PM, Shane Kamar <sha...@ma...> wrote: > Thank you for your timely reply! > > I am indeed a Mac user; however, by the miracle of bootcamp and parallels, > I develop also for Windows (7) as well as POSIX (suseLinux). > > I personally love Mac--as well as the user interface conventions that Apple > has pioneered--and would love to further streamline UBC inside and out to > help it run smoothly with an interface equally beautiful across platforms; > that being said, for the best integration, a few forks on the G/UI level may > be necessary, as each platform offers an abundance of system integration > methods not exposed in QT. > > For example, Windows Vista introduced Windows Presentation Foundation, > which streamlines the creation of UI's in the style of the native (Windows) > operating system; with Windows 7 came several new elements not available > across platforms, such as Taskbar options, progress displays, etc. > > Equally, KDE and GNOME carry their own user interface elements, some of > which are exclusive their respective environments. > > Naturally, Mac OS X has been a study in user-interface ingenuity in the > eyes of developers and design critics alike. The latest iterations of the > popular Apple OS have brought stylish components such as Quartz--a graphical > programming language for laying out fast, runtime-rendered, smooth graphical > interfaces--as well as Core Graphics--a similar framework for > Objective-C--and various integrations with the operating system. > > With the growing movement of processing power from desktop (laptop) > computers to scalable server environments, I would just like to venture > forth the idea of having a host-able (server-run/headless) distribution of > UBC alongside the current desktop version. > > I have taken a precursory glance at the distribution as well as the code. > Being a mac/posix/windows user, I think it would be advantageous to set up > an automated cross-compiling build system, or failing that, an automated > distribution system that would compile appropriate redistributable s for > each appropriate platform. > > Also, I realize I am new to this, so I hope my ideas aren't taken as > arrogance, but I just though I would mention a thought I had on plugins. I > like your XML plugin system, and I think it would make it ever more flexible > if we were to implement a kind of "bundle" or "container" system for > plugins; Mac OS X marks certain folders as "bundles" and they are thusly > treated as a single file. Perhaps we could do something similar in the way > that Apple handles documents and iOS applications. > > An iOS application is just a folder containing everything that the app > needs to run, all zipped up with a manifest. > > Perhaps for the plugin system, we could do something similar: each plugin > could be a ZIP file containing: 1) executables for the appropriate tool(s) > needed for that plugin or group thereof; 2) the manifest (XML, but with type > declarations; 3) any (multi-platform) DLL or .so or .dylib that could be > used to plug-in the the main program. This would allow great flexibility > across platforms while assuming minimal requirements on the client-side. > > I hope these thoughts prove useful, and I very much look forward to working > with you on this wonderful project. Please let me know what your ideas on my > suggestions and observations. > > Unfortunately it's time for Linear Algebra so I must send this email now > and rush to class. > > I will further peruse the source code after my 2PM class (it's 1PM here), > so I will be sending another email shortly thereafter. > > Thanks again, and best wishes, > > Shane Kamar > > On Apr 8, 2011, at 11:36 AM, Me wrote: > > Shane, > > We would love to have you. In particular, based on your email I'd venture > to guess that you are a mac user. We have been trying to work with mac in > mind, but none of the developers have a system running OSX, so we've had to > rely on information from a couple people who tried running it once. If you > wouldn't mind ensuring we have a good experience running on OSX, that would > be wonderful. > > > Just a quick overview of the project so far: > The other active developers and myself are undergraduate students at > Rensselaer Polytechnic Institute. > We are developing with the Rensselaer Center for Open Source > http://rcos.rpi.edu. > > Git usage: > As far as development goes, we are still in fairly early stages of > development. > We develop new features in separate branches. These can be pushed to the > sourceforge server for collaboration. > We have two main branches, devel and master. > Our policy is that the devel branch should compile, without major breakage > if possible. > The master branch should be usable, so that building it produces a working > program. > When we have a stable program with new features, we want to release. > So far we've been pretty flexible about sticking to these requirements, > both because we've been trying to get our program to the stage of being > ready for general use, and producing demo code to get our program out in > the wild. > > Previously, we have done a lot of planning offline, as we have face to face > interaction on a regular > basis. We will make an effort to get our discussion online in order to keep > you in the loop. > We have a small amount of documentation on our wiki and bug tracker, but it > is incomplete. > I also built doxygen documentation from the source code, although it is out > of date because currently > I build and upload it manually. We also have a simple task list on our > website, which makes it easy > to list what we're currently working on. Again, we haven't been as good > about keeping them up to date > as we should be because we meet face to face, but we will push to get our > work online. > > To get started setting up a development environment, visit: > > http://sourceforge.net/apps/mediawiki/unibatchconv/index.php?title=Developer_Documentation > > Post questions you have about the code on the forum (which we can start > using finally): > http://sourceforge.net/projects/unibatchconv/forums/forum/1231106 > > Just so you are aware of our contributions: > I'm einarin, I developed Converter and the XML converter specification. > Alex Hunt (FudgeMutator) developed the conversion management code and most > of BackendManager. > Frank Kotarski (Kotarf) developed the multistage conversion code. > Jon Kriss (Jonnyk127) developed the GUI. > > Feel free to email me with any questions. I look forward to working with > you. > > John Dickinson (einarin) > > > Shane Kamar > Sophomore, Auburn University Honors College > CSSE, College of Engineering > <sc...@au...> > (256) 682-2032 (mobile) > > > Shane Kamar Sophomore, Auburn University Honors College CSSE, College of Engineering <sc...@au...> (256) 682-2032 (mobile) |
|
From: Me <kc...@gm...> - 2011-04-10 18:50:02
|
Shane, It looks like you have a lot of great ideas. Sorry I didn't respond earlier, I had a bit of a crazy end to the week. I want to include the other developers in the discussion, so I CC'd the mailing list. I just set it up, but I think email might be a good way to communicate. Does that work well for you? I think I prefer it to the forum for development discussion. Sourceforge's forum isn't exactly stellar anyway. As far as the GUI goes, currently we just have the Qt GUI. It was chosen because it is pretty functional on Windows/Linux and works with OSX, unlike GTK. Delving into the code a bit, currently we have some Qt code in BackendManager. This code primarily serves to send debugging information to the GUI. I want to refactor all of that code out into it's own class. I was thinking something along the lines of a generic logger class, with our current Qt code inheriting that as an implementation. Once that code has been pulled out, we should be able to build the backend code as a standalone library that only depends on Boost. I think it is important to have the backend separate so that it can be used as a converter engine for any program. I was waiting to implement that because Alex is currently implemented multithreaded loading in BackendManager, which involves some significant changes to it's operation. Once the backend code is no longer explicitly tied to the frontend, we will be able to do things like write alternate UI's. I have done some work on a command-line interface, which I'm using to provide a test framework for cleaning up the backend Qt dependency. As far as the plugin system goes, one concern we had was patent issues. I don't think we can legally distribute LAME binaries without owing licensing fees. However, some sort of system like you suggest would be excellent for providing drop-in format support, particularly on Window and Mac. My goal with the XML converter description was to avoid having to write any code to support a new converter, so hopefully we won't have to worry about providing a cross platform library in any package. Your suggestion of a server-based converter is interesting, I hadn't thought of that. However, I think we need to work on core functionality first, as there are several areas where we are still lacking as a converter program. Once that is done, it would be very cool to be able to support "cloud converting" or similar. We already have a cross-platform build system, CMake. I don't know if it supports cross-compiling. Setting up a build server to automatically build releases should be doable, although I'm not sure that our code has reached a maturity level to make that worthwhile yet. I'm glad you have so many great ideas. Do you have a development environment set up yet? Let me know if there are any issues I or the other developers can help you with. Think about picking one improvement to work on, and starting from there. If there is anything that isn't clear in the code, don't hesitate to ask. Thanks, Johnny Dickinson On Fri, Apr 8, 2011 at 1:59 PM, Shane Kamar <sha...@ma...> wrote: > Thank you for your timely reply! > > I am indeed a Mac user; however, by the miracle of bootcamp and parallels, > I develop also for Windows (7) as well as POSIX (suseLinux). > > I personally love Mac--as well as the user interface conventions that Apple > has pioneered--and would love to further streamline UBC inside and out to > help it run smoothly with an interface equally beautiful across platforms; > that being said, for the best integration, a few forks on the G/UI level may > be necessary, as each platform offers an abundance of system integration > methods not exposed in QT. > > For example, Windows Vista introduced Windows Presentation Foundation, > which streamlines the creation of UI's in the style of the native (Windows) > operating system; with Windows 7 came several new elements not available > across platforms, such as Taskbar options, progress displays, etc. > > Equally, KDE and GNOME carry their own user interface elements, some of > which are exclusive their respective environments. > > Naturally, Mac OS X has been a study in user-interface ingenuity in the > eyes of developers and design critics alike. The latest iterations of the > popular Apple OS have brought stylish components such as Quartz--a graphical > programming language for laying out fast, runtime-rendered, smooth graphical > interfaces--as well as Core Graphics--a similar framework for > Objective-C--and various integrations with the operating system. > > With the growing movement of processing power from desktop (laptop) > computers to scalable server environments, I would just like to venture > forth the idea of having a host-able (server-run/headless) distribution of > UBC alongside the current desktop version. > > I have taken a precursory glance at the distribution as well as the code. > Being a mac/posix/windows user, I think it would be advantageous to set up > an automated cross-compiling build system, or failing that, an automated > distribution system that would compile appropriate redistributable s for > each appropriate platform. > > Also, I realize I am new to this, so I hope my ideas aren't taken as > arrogance, but I just though I would mention a thought I had on plugins. I > like your XML plugin system, and I think it would make it ever more flexible > if we were to implement a kind of "bundle" or "container" system for > plugins; Mac OS X marks certain folders as "bundles" and they are thusly > treated as a single file. Perhaps we could do something similar in the way > that Apple handles documents and iOS applications. > > An iOS application is just a folder containing everything that the app > needs to run, all zipped up with a manifest. > > Perhaps for the plugin system, we could do something similar: each plugin > could be a ZIP file containing: 1) executables for the appropriate tool(s) > needed for that plugin or group thereof; 2) the manifest (XML, but with type > declarations; 3) any (multi-platform) DLL or .so or .dylib that could be > used to plug-in the the main program. This would allow great flexibility > across platforms while assuming minimal requirements on the client-side. > > I hope these thoughts prove useful, and I very much look forward to working > with you on this wonderful project. Please let me know what your ideas on my > suggestions and observations. > > Unfortunately it's time for Linear Algebra so I must send this email now > and rush to class. > > I will further peruse the source code after my 2PM class (it's 1PM here), > so I will be sending another email shortly thereafter. > > Thanks again, and best wishes, > > Shane Kamar > > On Apr 8, 2011, at 11:36 AM, Me wrote: > > Shane, > > We would love to have you. In particular, based on your email I'd venture > to guess that you are a mac user. We have been trying to work with mac in > mind, but none of the developers have a system running OSX, so we've had to > rely on information from a couple people who tried running it once. If you > wouldn't mind ensuring we have a good experience running on OSX, that would > be wonderful. > > > Just a quick overview of the project so far: > The other active developers and myself are undergraduate students at > Rensselaer Polytechnic Institute. > We are developing with the Rensselaer Center for Open Source > http://rcos.rpi.edu. > > Git usage: > As far as development goes, we are still in fairly early stages of > development. > We develop new features in separate branches. These can be pushed to the > sourceforge server for collaboration. > We have two main branches, devel and master. > Our policy is that the devel branch should compile, without major breakage > if possible. > The master branch should be usable, so that building it produces a working > program. > When we have a stable program with new features, we want to release. > So far we've been pretty flexible about sticking to these requirements, > both because we've been trying to get our program to the stage of being > ready for general use, and producing demo code to get our program out in > the wild. > > Previously, we have done a lot of planning offline, as we have face to face > interaction on a regular > basis. We will make an effort to get our discussion online in order to keep > you in the loop. > We have a small amount of documentation on our wiki and bug tracker, but it > is incomplete. > I also built doxygen documentation from the source code, although it is out > of date because currently > I build and upload it manually. We also have a simple task list on our > website, which makes it easy > to list what we're currently working on. Again, we haven't been as good > about keeping them up to date > as we should be because we meet face to face, but we will push to get our > work online. > > To get started setting up a development environment, visit: > > http://sourceforge.net/apps/mediawiki/unibatchconv/index.php?title=Developer_Documentation > > Post questions you have about the code on the forum (which we can start > using finally): > http://sourceforge.net/projects/unibatchconv/forums/forum/1231106 > > Just so you are aware of our contributions: > I'm einarin, I developed Converter and the XML converter specification. > Alex Hunt (FudgeMutator) developed the conversion management code and most > of BackendManager. > Frank Kotarski (Kotarf) developed the multistage conversion code. > Jon Kriss (Jonnyk127) developed the GUI. > > Feel free to email me with any questions. I look forward to working with > you. > > John Dickinson (einarin) > > > Shane Kamar > Sophomore, Auburn University Honors College > CSSE, College of Engineering > <sc...@au...> > (256) 682-2032 (mobile) > > > |
|
From: Me <kc...@gm...> - 2011-04-08 08:51:02
|
Testing to make sure the mailing list works. Johnny |