Yesterday morning, I emerge-ed kdebase, a basic KDE package including most things a KDE user might want to have.

Compilation started at about 10am.

And ended this morning at around 5am.

That's 19 hours of compiling!!!

I'm nowhere near the physical machine now, so I'll fiddle with this soon and report back here...

Cool thing : update does not mean "I'll crush your local files with my version, foo'!" (think of Mr T)

It seems to mean : I know what version you last synchronized with, let me download the latest patches since then and apply it for you.

Which means you don't loose your changes. Which means it's all coo'.

Anyone who'se had experience with version management is probably guessing that I'm toying with multiple users "checking out" the same file.

See, using SourceSafe, we didn't allow that. Single check outs. You want it? Someone else got it? Tough. We don't trust code that'll try to "merge the changes".

With CVS, it seems we don't have such an option and we're reluctantly trusting the merging mechanism.

And he's a cool one.

I'm starting to like CVS, but I'm still not comfortable with it. So I just might try and give that "subversion" fella a try.

Well, I would do it now, except that my little padawan of a server is having fun emerge-ing kdebase, which consists of about 15 packages related to KDE (including KDE itself and Konqueror).

It's been compiling happily for the last... uhm... 6 or 7 hours. Maybe it'll be done this week? Then I'll check into subversion. Don't we just looove that name?

It just drips with cool. It feels like the dark side... with a server named Anakin, what could be best?

Since I got me thinking about Mr T, here's a quote from the man:

When I was growing up, my family was so poor we couldn't afford to pay attention.

Okay, I think I got it.

Its freakish, but I got it.

Commit sends your version to the server. Updating sends the server's version to your local site synchronizing tries to to both at the same time.

The thing you need to know about Eclipse, that might save you headaches, is that, when synchronizing the left window is the local version and the right window is the remote version. NOT THE OTHER WAY AROUND.

I think the windows are actually labeled when you're doing "every day" sycnchronizing... but not when you're doing the "I'm mapping this project to a CVS database for a first time" synchronizing. (My local version is not up to date for every file. Some files are more up to date than the remote database.)

A little labelling might've helped me to not fuck up my remote database.

Well, I caught on after having messed up only two files... (that's before I figured how I could've messed up a whole packages in one signe shot - or the whole project). So I said unto myself: "no problem. I'll just roll back to my previous version for these files. No biggie, right? right? RIGHT??

Well, I've still to figure ou the rollback thing. I figured I can tag a file to an older version (and can no longer update it). Ended up figuring out how to get rid of the friggin flag.

What I ended up doing is getting version 1.1 of the file and committing the changes. So now my files are up to 1.3. But 1.2 is useless and I would've wanted to thrash it.

If you're into Eclipse and CVS, help, enlightment and tips & tricks are welcome!!

There's a curve for me in CVS, and it's of the learning type.

See, I've been using MS Source Safe for years. Long enough for me not to remember how odd it was for my brain to wrap around its (it how seems) real simple concepts.

Now it's CVS's turn to do this to me.

Never mind that I had trouble getting it working. Never mind that I'm still not sure if I want to run its authentication as ssh or as pserver.

I got it working, I made it work with my IDE of choice (Eclipse!) and now I'm struggling to understand.

Gone are the well understood concepts of "checking in" and "checking out". Maybe it's Eclipse's fault? I've got update, commit, synchronize, branch, merge, tag as version, apply patch, create patch, etc.

Most of these I get. I read these and they all make sense.

But in practice, I haven't had that luck.

I succeeded (I think) in properly creating projects in my CVS database and sending my code. All fine.

Other people on my projects are still using SourceSafe (until I can teach'em how to use CVS efficiently - I don't want to lose productivity!). So as I uploaded a project on my pristine new CVS gig, I heard a new update was ready in SS (that's for Source Safe.)

So I "Get latest version"ed the project from SS (I don't know which files got updated), planning to send the updated stuff back to the CVS server. That's when I started not getting what was going on.

In SS, I know how to check out files without actually getting the file from the server. Idea was to check out my files from CVS, don't get the code, check in everything (hopefully only what had truly changed). That was the plan.

And I also wanted to test integration within Eclipse. So I went to one of the file I knew had been modified, but not "checked out" and typed. As expected, Eclipse checked out the file for me. What I didn't expect was for it not to get the latest version from the server. Maybe it figured the date on my file was more recent than the servers' ?

I could've just said screw it. Its doing what I want, let's just "check in" (which should now be also known as "commit", which also makes sense) the files and be done with it.

Due to my explorative nature and my wish to understand just what is going on, my objective now was : let's get the version as it stands from CVS first. Let's "check out" the whole CVS project, update it with SS and check it back it.

That's what I never managed to do. "Update" didn't do it (what does it do?). "Synchronize" wants me to merge the changes manually. No thank you.

I was pressed for time, so I went to just committing any changes CVS could find. Worked. But I'm not satsified with my gained knowledge.

Later, I'll try to sync to a remote project from home and update my local copy.

I must be insane.

I'm about to go to bed. Its 3am. I just finised to setup my CVS server on Anakin.

What can possess a guy to do that? To stay that late? I should've gone to bet at 10pm. Yesterday.

What took me so long? Well, many things. Understanding how CVS worked. Setting rights and groups "properly". Figuring out how "Eclipse" connects to CSV using SSH.

After much labor, I finally succeeded. I'm uploading a few pieces of code right now. Just because I can.

Now, working remote'll be much more cool.

I just need to get used to these tools... then try to make the rest of my friends use it too :P

I love this shit. It makes me feel like a young hacker again. When DOS was around and I was the only one on the block who knew what he was doing.

I feel like that kind of magician again.

Like how I felt when I built a small set of functions that allowed me to use the video card in "mode X". I was cool, some kinda genious kid. I was unique. I was all powerful. Nobody understood what I was doing - but they knew I knew what I was doing... I had results! They stood in awe at my might. Perplexed and amazed by what phenomenal abstractions my brains was able to harness.

Then progress came along and.. well... priorities shifted.

It's cool to reconnect with these feelings again :P