[Citadel Development] (no subject)
Hmm.. Windows installer, eh? Well, I guess I should look at this, then. I'm probably the grudging master of installing things on Windows.
[Citadel Development] (no subject)
Excellent. Then I shall step out of the way and let the master both build and install the application on Windows. Thu May 22 2008 05:27:02 AM EDT from [EMAIL PROTECTED] Hmm.. Windows installer, eh? Well, I guess I should look at this, then. I'm probably the grudging master of installing things on Windows.
[Citadel Development] (no subject)
Hrm... I really need to get my new computer soon, so I can install cygwin, build Citadel, test it, then create an installer for it. I use VirtualBox on my ubuntu desktop and then perform the builds under something as mundane as Windows 2000. In this way, the build should work on both 2000 and XP. Vista remains to be seen, but I expect that it should run there as well. Please - don't build it on Vista and then distribute the result. Vista sux more ass than anything that has ever come out of that god-forsaken Redmond company, and I really don't want to see some sort of error thrown at me because some part of an application requires it. I don't know if it would require something like this, but then again - it's Windows - who knows? I have some very nice software for building an installer. I can even build an updater that can automatically notify when updates are available. I use the freely available Visual Studio Installer. Granted, it does not have the auto-update feature, but that is secondary anyway. Besides, VSI is free. How are we going to tell when an update proposed for release has broken the build on a particular platform? Simple - you don't. Once an application has been installed, do not go and try to update it automatically. That's bad karma all the way around because your changing the end users system - even more so if this is going to reside in a psuedo-production space. It makes more sense to simply let the end user know that there is a new release available and then ask *them* to update to it just like ClamAV does with their Windows installer. At the same time, when the build master puts a release together, he/she person should try it out on as many platforms as possible. Test. Test. Test. Do we have some kind of automated testing system, for that matter, that tests low-level functions in some way? Nothing that I am aware of. I have not seen anything out there that relates to CygWin like this. If you were to spend the time, maybe Winbatch, or something similar could be used since it is in a web browser. Thu May 22 2008 05:42:25 AM EDT from [EMAIL PROTECTED] (FleebNet) Hrm... I really need to get my new computer soon, so I can install cygwin, build Citadel, test it, then create an installer for it. I have some very nice software for building an installer. I can even build an updater that can automatically notify when updates are available. But, if we're going to get into this degree of cross-platform development, I have a some questions: How are we going to tell when an update proposed for release has broken the build on a particular platform? Do we have some kind of build farm going on, that propagates failures upstream to some kind of collection point? Do we have some kind of automated testing system, for that matter, that tests low-level functions in some way? I wonder how hard it would be to build something like that.
[Citadel Development] (no subject)
Thu May 22 2008 07:51:04 EDT from [EMAIL PROTECTED] Hrm... I really need to get my new computer soon, so I can install cygwin, build Citadel, test it, then create an installer for it. I use VirtualBox on my ubuntu desktop and then perform the builds under something as mundane as Windows 2000. In this way, the build should work on both 2000 and XP. Vista remains to be seen, but I expect that it should run there as well. Please - don't build it on Vista and then distribute the result. Vista sux more ass than anything that has ever come out of that god-forsaken Redmond company, and I really don't want to see some sort of error thrown at me because some part of an application requires it. I don't know if it would require something like this, but then again - it's Windows - who knows? I am waiting on a WinXP Pro box. I'm not exactly a fan of Vista, either. This said, considering we have to use cygwin to build Citadel, which means using gcc, I strongly doubt it would matter if we built it on Vista, XP, 2000, or 9x. Even in a pure Windows development environment you can build for specific versions of Windows by setting the _WIN_NT #define to particular values (if I remember the macro name correctly). Ultimately, this isn't going to be a problem. I have some very nice software for building an installer. I can even build an updater that can automatically notify when updates are available. I use the freely available Visual Studio Installer. Granted, it does not have the auto-update feature, but that is secondary anyway. Besides, VSI is free. Ugh... yeah, that's an option, but man it's uncomfortable to use, if my memory serves me right. If one wants to stick with freeware, one can use nix and do a far easier job of it. It's flexible, free, but very text-based. I'm planning on using Advanced Installer. They have a free version available, although I actually bought a license for it, as I would like to play with the graphics, amongst other things (e.g. I'd like, if possible, to let the user select where to install things). How are we going to tell when an update proposed for release has broken the build on a particular platform? Simple - you don't. Once an application has been installed, do not go and try to update it automatically. That's bad karma all the way around because your changing the end users system - even more so if this is going to reside in a psuedo-production space. It makes more sense to simply let the end user know that there is a new release available and then ask *them* to update to it just like ClamAV does with their Windows installer. At the same time, when the build master puts a release together, he/she person should try it out on as many platforms as possible. Test. Test. Test. I'm asking the question in a broad way, actually... I probably should have asked in a separate message. People use Citadel on FreeBSD, various forms of Linux, and Macs, as well as possibly other environments... and now we're looking to run it on Win32. It would be nice if we could guarantee that it at least compiles on each of these platforms. To that end, it would be nice if we had some kind of system that could identify compile failures (at least) to give developers a heads-up when they've implemented a change that broke on another platform. Citadel is growing, and it's getting big enough that such considerations probably need to be made. I know other OpenSource projects have this problem. In particular, Firefox. Might want to check that out.
[Citadel Development] Citadel commit log: revision 6307
r6307 | ajc | 2008-05-22 10:36:37 -0400 (Thu, 22 May 2008) | 1 line Changed paths: M /trunk/citadel/modules/openid/serv_openid_rp.c M /trunk/citadel/server.h Fetch assoc_handles and use them in checkid_setup requests
[Citadel Development] Citadel commit log: revision 6308
r6308 | ajc | 2008-05-22 12:45:07 -0400 (Thu, 22 May 2008) | 1 line Changed paths: M /trunk/citadel/modules/openid/serv_openid_rp.c M /trunk/webcit/auth.c M /trunk/webcit/webcit.c M /trunk/webcit/webcit.h More work on OpenID 1.1 support
[Citadel Development] (no subject)
I'm planning a nightly build system on one of my servers here. The plan is to write a script that checks subversion for new code, grabs the current head and then builds it something like easyinstall does now. I'd then have build errors posted to a special room on Uncensored. I think I could do 3 different environments, maybe 4. Centos 4 Centos 5 Debian and possibly fc8 but that is a laptop so its not always on. Getting round to writing the script and setting up the cron job would be the hold up Wait I think we have half of this already. QueasyInstall IG, What code level does QueasyInstall provide? If its not current svn head can we get something like Queasy that is SVN head, then I'd just need a script to grab the queasy script and run it, then parse over the build log and post errors.
[Citadel Development] (no subject)
The plan is to write a script that checks subversion for new code, grabs the current head and then builds it something like easyinstall does now. Before we go reinventing the wheel here, has anyone looked at Tinderbox? http://www.mozilla.org/tinderbox.html
[Citadel Development] (no subject)
Do Mai 22 2008 05:42:25 EDT von [EMAIL PROTECTED] (FleebNet) Do we have some kind of automated testing system, for that matter, that tests low-level functions in some way? I wonder how hard it would be to build something like that. So you want some sort of tinderbox2 with some VMWares? I Think a real port would be if one would try to compile with mingw...
[Citadel Development] Citadel commit log: revision 6312
r6312 | ajc | 2008-05-22 15:30:15 -0400 (Thu, 22 May 2008) | 1 line Changed paths: M /trunk/citadel/modules/openid/serv_openid_rp.c Made the assoc_handle field bigger, but we may still have a problem
[Citadel Development] Citadel commit log: revision 6313
r6313 | ajc | 2008-05-22 16:08:59 -0400 (Thu, 22 May 2008) | 1 line Changed paths: M /trunk/citadel/modules/openid/serv_openid_rp.c hmmph. adjusted string lengths again
[Citadel Development] (no subject)
Thu May 22 18:15:37 2008 from [EMAIL PROTECTED] (Uncensored) Do Mai 22 2008 05:42:25 EDT von [EMAIL PROTECTED] (FleebNet) Do we have some kind of automated testing system, for that matter, that tests low-level functions in some way?I wonder how hard it would be to build something like that. So you want some sort of tinderbox2 with some VMWares?I Think a real port would be if one would try to compile with mingw... VMWare might be the trick... that or a variety of people with the various boxes.I tried using Microsoft's stupid Unix Compatibility Layer' thing, whatever it's called. Total crap. Maybe Mingw could do it, being closer to cygwin. Dunno.
[Citadel Development] Citadel commit log: revision 6315
r6315 | ajc | 2008-05-22 23:02:09 -0400 (Thu, 22 May 2008) | 1 line Changed paths: M /trunk/citadel/modules/openid/serv_openid_rp.c M /trunk/webcit/auth.c AOL's OpenID implementation sucks.
