My apologies for the delay in getting back to this thread. I have been exceptionally busy the past couple of weeks, as my wife went in to the doctors office and got nuked. Right now, I am playing Mr. Mom to my three kids and we're waiting for her return tomorrow evening. Cancer is a booger, no matter what anyone says. We got lucky, and after they removed her thyroid, found that it's all in one place so they can cure it.
Anyway, with that being said... I like the approach... I really do. It makes sense. Cygwin is a "compatibility layer" for Winchunks, or at least that is what they say. I've noticed in some of the other apps that I have ported over that there have been minor issues here and there. If we can update the installer to account for it, that would be wonderful. I've noticed, however, that the version of iCal does not want to compile correctly. To get around the issue, I used the .31 version. I left the versions of Expat, and LibSieve alone. While I was at it, I updated the version of Berkley from 4.3.29.NC to 4.6.21. I can keep up with the source tree for the citadel server itself, and libcitadel but not for webcit. For some strange reason, there is an issue with the version that is there. Anything past version .24 simply does not work. I think that I have posted here and there about this before. Essentially, there is a bug somewhere in the updates past the .24 version in which webcit will establish a connection with the browser, and then open up a connection to the citserver, and the citserver will see the connection come in, but no response is sent back up the chain. I've dropped WireShark on the connection and can see the request go through, but no response. For what it is worth, the FreeBSD compile has the same bug. I've got a successful build on that platform as well and will be uploading it to my site within the next couple of days. If you would like a FreeBSD 6.2 box to play around on, I can certainly setup an SSH account for you on one of mine. At least that way, you can try out the build and see the bug that I have been talking about. A Windows service? Hey, sounds good to me... I am still working on the installer. From within the installer, I can execute additional programs to setup connections, shortcuts, and service entries. That would make it really easy for Winblows users to install the app. >Mon Apr 07 2008 11:37:50 AM EDT from IGnatius T [EMAIL PROTECTED] > >Thoughts on the maintenance of Windows packages... > > >> >How should we handle compile bugs, or bugs which seem to work their way >>out >> >into view during testing? Likewise, I am using Cygwin to do the builds >>since >> >it is a common tool available for this sort of activity. The code, >>therefore, >> >is the very same, unmodifed source that is traditionally used on a *nix >> >platform. >> > >My initial thought is that we should simply consider Cygwin to be just "one >more flavor of unix" upon which we have to test. If something breaks in >that build, we work around it, just as we would work around other >problematic unix flavors such as OpenBSD. > >Is it safe to guess that you are at least keeping up with the source tree >in svn? If not, that would be the first place to start. As the port >progresses, I'd be very happy if you could figure out a way to get Citadel >and WebCit to be installed as true Windows services. > >The >setup programs already have the ability to offer several different front >ends for user dialogs. If you want, we can add a Windows mode for that too. > >
