Casper Hornstrup wrote:
I hope the parties involved don't take this personally, but I have to speak my mind about something I think we're doing terribly wrong and as a result is hurting the project.
And don't take my replies personally either...
Basicly we have the following situation (and please correct me if I'm wrong).
Hartmut spends some time examining the registry and object manager problems.
I think Thomas deserves credit for this as well.
Alex has partially rewritten the object manager to fix the problems, but can't get get further because he needs to wait for Thomas to finish his handle table implementation.
Not quite true... it is blocking me, but so are at least hundreds of lines that I have to fix and review and re-edit in my rewrite, which is at LEAST more then a month away, just like the Cc rewrite has/will be in progress for a long time.
Meanwhile, Thomas seems to be doing everything but finishing his handle table implementation.
Thomas has been testing this for over a week and he almost gets the GUI to boot...it's unfair to accuse him of slacking off.
Alex keeps an increasing number of bugfixes (not only bugfixes for the object manager) on his private miscelanea branch and thus keeping these bugfixes away from 25 developers and whatever number of testers we have.
"Private"? there's no password on it, and anyone is free to test it, and this is exactly what they are doing: 1) Thomas - Added many features, fixed some bugs and alerted me to some issues 2) Steven - Tested the branch and found out his laptop now works, helped me with some debug issues 3) Jim - Tested the branch extensively almost every day for HOURS. 4) Art - Did lots of testing with network applications and the stack, noted speedups and improvements.
These are only some of the people that have helped and tried out and shared code and hints. You keep telling people to use a branch, and now you accuse me of keeping it "private". The reason I've done it in a branch is because I was getting sick of people saying how I broke HEAD. And if this branch would be in head since the start, I would've broken it a large number of times. The existence of my branch has done nothing to hinder development on HEAD. The only person who needed its features was Thomas, and he has been working with it.
The 25 developers and testers will have to live with these bugs until the branch is merged to trunk which appears to be several weeks away.
Just like we all have to live without Hartmut's fixes to Cc, Filip's DMA Patch, Eric's RPC stuff, Art's Ext2Fs support, and the numerous other local fixes the developers have made and keep for themselves until they are ready to test. Unlike them however, I have made my stuff PUBLIC so that others could help out (and they have). I am the last person you should blame for "keeping stuff private" and making devs live with bugs. My branch is not weeks away...it has been ready since Monday, and I've been waiting on Steven to prepare for 0.2.6, which he told me he was going to do Wednesday "OUT OF LACK OF TIME". I find it extremly disappointing that Steven says that 0.2.6 was delayed to today because of my branch... It was delayed a week, yes, but that was the result of people REQUESTING it after I asked on the ML.
Hartmut seems to be willing to fix the object manager and registry manager bugs on trunk, but if he now did, he would be duplicating work.
No he wouldn't, he would do everyone a good favor. My object manager works completely different from ros's, so it wouldnt' be code duplication.
If this is what has happened, then we should make sure it doesn't happen again so we don't waste our (very) limited resources. If this is not what happened then I apologise for wasting bandwidth on this mail.
Casper
Once again, I hope nobody takes this personally, but I wanted to clear up the situation.
Best regards, Alex Ionescu