On Thursday 19 April 2007 23:26, Robert Nelson wrote: > Comments inline below > > > -----Original Message----- > > From: [EMAIL PROTECTED] [mailto:bacula-devel- > > [EMAIL PROTECTED] On Behalf Of Kern Sibbald > > Sent: Thursday, April 19, 2007 3:10 AM > > To: bacula-devel; bacula-users > > Cc: Shane Coughlan > > Subject: [Bacula-devel] Bacula Status -- change of direction for > > myparticipation in the project > > > > Hello, > > > > Open Source is a fantastic success story, and shows every sign of becoming > > a > > gigantic snow ball over the next few years. Usage of Bacula is increasing > > significantly, which is very pleasing. However, the development side of > > Bacula, with one or two exceptions, I consider a total failure. > > > > Bacula has received quite a number of submissions other than my > > contributions > > over the years. However, in general, these submissions have been made > > without documentation (leaving it to me to document) and the developer > > after > > a short time has for various reasons moved on to other things (change of > > job, > > change of life status, other interests, ...). All this is normal, but > > what I > > find very disappointing is that with only a couple of exceptions that come > > to > > mind there are no permanent Bacula developers other than myself. (Just so > > that there is no confusion or ill feelings, Scott, Dan, and Eric have been > > around for quite a while and have and are making significant contributions > > -- > > there are certainly others who don't come to mind immediately, so, please > > accept my appologies -- also, I think the users list is working quite > > well. > > What I am talking about is the lack of highly qualified, committed, > > and "permanent" Bacula developers). > > > > I hope I'm not one of the developers that you feel isn't qualified or > committed. I'm not sure what you mean by "permanent".
I think you took my statements from the wrong angle. I wasn't complaining about any of the existing developers, but indicating that the Bacula project does not have sufficient qualified, committed, and permanent (long term) developers. I gave an idication somewhat above by what I mean by permanent. In your case, you are certainly qualified, probably the most technically qualified of all the people who have worked on Bacula. You have not been around long enough for me to consider you "permanent". Only time will tell. However your participation, which is very much appreciated seems to come in bursts rather than consistently day after day. This is fine, as I will accept all the help I can get in any form. Example, your last email was on 6 March. > > I answer the questions on the mailing list that are specific to the Windows > version. I usually ignore those that are general Bacula questions, even if > they are running Windows, unless there is some Windows specific aspect. That's fine. > > > > My point here, is that today, I am essentially the only developer doing > > documentation, the only developer fixing bugs, the only developer doing > > testing on the different platforms. After 7+ years of doing it 12-14 > > hours a > > day for 6 and sometimes 7 days a week, I am more than a bit disappointed > > that > > there isn't more help. The fault may possibly be my own, but what ever > > the > > case may be that is where we are. > > > > I have also been fixing bugs. Yes, you have fixed bugs, and I appreciate that. However, I don't consider you as an active developer that is doing bug fixing. I will give you a few examples, and they aren't at all intended to criticize you, but to indicate the level of committment that I see in your participation (which is of course for you to define) and thus indicate to you more clearly why I made the Win32 decision. The last time you made a commit was 15 March over a month ago, and before that it was 7 Feb. This is fine, and I am happy to get those commits/fixes. > > > As an example of what I am lamenting here is that there is a Win 2003 bug > > open > > since 20 March where restore of encrypted (and compressed if I remember > > right) data fails. Another example is that despite my repeated requests > > over > > something like a six month period, no one (at this moment) has "signed up" > > to > > do Win32, Solaris, or FreeBSD regression testing. > > > > As far as this specific bug is concerned, I just realized it is assigned to > me. However I thought there was someone else working on the Encryption > code. Even though this problem is reported on Windows the problem is more > likely to be a generic problem with encryption support. This bug is a good example of what I am complaining about. The encryption code was originally written by another programmer, but the last serious bug was fixed by me (significantly after notifying the programmer). Logically, as you mention, it should be the original programmer that fixes it, but you are already aware that does not work well in this case, because like me, you have previously identified and fixed a major encryption (actually decryption bug). Concerning specifically the bug report # 807 in question. I consider it a critical bug (the worst kind) because a user is not able to restore data that he wrote. It may not be Win32 specific, but on first view, I am assuming it is. The bug report was submitted the 20th of March, and as far as I can tell, there has been no progress in resolving it. Contrary to what you say above, you did know that it was assigned to you, because you responded to the bug the 20th of March, and presumably you assigned it to yourself or the system had already done so. I want to make clear that I am not complaining about either the original encryption programmer nor about you. This is Open Software, so you are totally free to respond or not. However, having a critical bug open against a Win32 (server if I am not mistaken), is not something that I can support/fix, and given the time the bug has been open, the level of support for the Win32 does not live up to my standards, thus I have decided to drop support for it rather than provide a product that users believe is supported and then provide no support. Should the support level improve in the future, perhaps we can support Win32, or perhaps I will find commercial users who have a corporate interest in having it supported, then things will change. > > > My solution to this problem is several fold: > > > > 1. As of release 2.2.0, Win32, Solaris, and FreeBSD will no longer > > be "officially" supported platforms. So that there is no > > misunderstanding, I > > definitely would like to see them supported, but I am no longer going to > > do > > it alone, which means that I will accept patches for them, I will not > > accept > > bug reports that are specific to those platforms, they will be documented > > as "use if it works for you but don't complain if it doesn't", and the > > project will no longer supply binaries for the Director and Storage daemon > > for Win32. Users can build it themselves if they want to use it. > > > > I don't understand why you are taking this position in regards to the > Windows version. We've talked a number of times about the support and > documentation issues. I agreed to provide support for any Windows specific > problems which I have been doing. Thanks for your support. I sincerely hope you continue to provide it. However, as I wrote above what you are currently doing is not sufficient for Win32 (other than the client) to be officially supported by Bacula at the moment. > As far as documentation is concerned, you > said you would make the changes if I provided a description of the > differences. In fact, in the last email you seemed quite happy when I > mentioned that there aren't any real configuration differences other than > the device names. Yes, what you write is correct, but it isn't the full story. I had previously asked you for documentation and when you said you were not fluent in LaTeX I offered to take ASCII and convert it to LaTeX. Then when no text was forth coming, I (stupidly) offered to write it myself. Rarely, but sometimes, I do go back on a promise to do something. This is one such case. I've decided that we do not support Win32 server nor will we ship the binaries until we can support it in the way that the project has always done in the past: 1. have full documentation (there is none in the manual on the Win32 specific parts of the server). 2. Have a person willing to look at and find solutions to serious bugs so that the bug doesn't sit in limbo for one month. Again, I really want to emphasize that I am not criticizing you or any other developer. Developers are free to come and go and choose what they want to work on. You are not getting paid to work on Bacula (nor am I). However, if I cannot find the people to do the documentation, nor to fix the bugs, then I think it is reasonable to provide the code on an "as is" basis without officially supporting it. > > The README file included with the installation documents all the differences > of which I'm aware. The README needs to be integrated into the manual -- I haven't looked at it so I am not sure what parts are appropriate for inclusion in the manual, and I suspect the manual needs a lot more things, such as most of the information that you have provided in emails for debugging and using your new Win32 utilities. To get this all in the manual is not a non-trivial amount of work. > > I periodically run the Windows regression tests (which I ported). You've > also asked for help running the regression tests which I agreed to do. But > each time you've indicated it would be at some point in the future but > you've never got back to me that you were ready. Please don't put this on my back. I sent an email a week or two ago saying that I needed testing for Win32, Solaris, and FreeBSD for the current SVN. I got back no response, thus my decision to "unsupport" those products until we have a resolution. It seems likely that there is now one for FreeBSD -- I have received back indication that all the tests run for FreeBSD, and we are moving forward defining a specific procedure. I have not received such confirmation for Solaris nor Win32. > > I also offered to do the Windows releases for you and you agreed it would > help. However, whenever you are making a release you've said that you were > taking care of the Windows release. Yes, what you write here is true. Since I run the Win32 client, I do build the binaries for release (as I package the source for release), and I do run the Win32 client in production. That is not a problem for me, and I don't have a problem continuing to do it. I also would not have a problem transferring it to someone else, but it isn't really necessary at this moment. Be careful in the terminology. I am happy to create a binary Win32 package because it is done 100% on my Linux machine, but the term "take care of the Windows release" implies a lot more (documentation, support, ...). And it is the documentation and support that I consider for the moment insufficient. > > Currently I'm working on the new Bacula Tray application. This will replace > the built-in tray code. It will eliminate the need to run using the SYSTEM > account. It also allows remote systems to be monitored. Encryption (TLS) > is supported ensuring secure password and data exchange between the monitor > application and the remote servers. Great. > > I believe this should eliminate four of the most common Windows issues / > requests: > > Ability to run using account other than SYSTEM I assume you are referring only to the tray monitor and not the Client. > > Restrict users from stopping the service from the tray application. > > Access the monitor application from non-console remote sessions. > > Encrypted sessions between the monitor application and the servers > for use over the internet Great. > > > 2. I previously suggested creating a formal Bacula Open Source project or > > a > > foundation, but I have totally dropped that idea due to lack of community > > participation. > > > > The only requests related to this have been the signed copyright > assignments. If there is something else you expect from us, please let us > know. I think you misunderstood what I was referring to. You have done everything that is needed for the copyright assignment -- that is no problem. You personally don't need to do anything more than you are doing, unless you want Win32 server officially supported, and I think I've explained that above. What I meant by item 2 was that I think the Bacula Open Source project (as opposed to my efforts to help 3rd parties to provide Bacula support) should be more officially organized along the lines of other large projects, with a "charter" and a "governing board", ... However, I've dropped that project -- at the moment it is not something important for the Bacula project. Best regards, Kern > > > 3. As of today, the gnome-console (renamed bgnome-console) and wx-console > > (renamed bwx-console) are deprecated and no longer supported by me. If > > someone else wants to pick up support of them, I'll be very happy to > > accept > > patches. > > > > 4. I am no longer personally going to maintain the projects list (Feature > > Requests). If someone wants to pick up maintaining it including the > > voting, > > I would be very happy. Obviously I'll continue to work on projects that > > personally interest me. > > > > 5. I will be devoting more of my time to a project that I previously > > mentioned > > that will provide training and support for Bacula engineers and for 3rd > > party > > Bacula Service organizations. The support provided will not be direct > > customer support but certification and level 3 support for professional > > service companies with the goal of promoting Bacula usage and code > > submissions. > > > > The concept here is that I am convinced that commercial organizations want > > to > > use Bacula (many use Solaris, so it is critical for them) but are hindered > > by > > the lack of qualified professional service. By insuring professional > > services for Bacula, I believe that we can compensate for the lack of > > commuity participation in the Bacula development process. This is because > > commerical/governmental/educational organizations will make significant > > contributions to the project when they have professional support. > > > > Best regards, > > > > Kern > > > > > > > > ------------------------------------------------------------------------- > > This SF.net email is sponsored by DB2 Express > > Download DB2 Express C - the FREE version of DB2 express and take > > control of your XML. No limits. Just data. Click to get it now. > > http://sourceforge.net/powerbar/db2/ > > _______________________________________________ > > Bacula-devel mailing list > > [EMAIL PROTECTED] > > https://lists.sourceforge.net/lists/listinfo/bacula-devel > > > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by DB2 Express > Download DB2 Express C - the FREE version of DB2 express and take > control of your XML. No limits. Just data. Click to get it now. > http://sourceforge.net/powerbar/db2/ > _______________________________________________ > Bacula-devel mailing list > [EMAIL PROTECTED] > https://lists.sourceforge.net/lists/listinfo/bacula-devel > ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
