One annoying thing in the CVS development...

[mark]

Is there a recent CVS-Version that cleanly compiles? Last time I checked it out, it did compile with errors and my app didn't work because of that.
So I waited one day and checked out again. Ah, that time CS worked fine...but CEL didn't compile anymore because of changes in CS. Then I saw a message that there are errors fixed in CEL depending on the recent changes in CS, so after 5 hours I checked out CEL with these fixes.
But CEL needs a "recent" version from CS so I checked out CS again... and now CS doesn't compile because of further changes...

I think I can do this update-and-compile-game for the next weeks without getting a working CS+CEL combination, because I saw many "I rewrote this" "I rewrote that" "Proposal for" messages in the mailinglist.


Most projects only checkin code that compiles. It doesn't have to work, but it does have to compile at least.
Don't you make a compile-test after changing something? I would never checkin code that doesn't compile, that would keep my team off from work.

And when you make severe changes like this SCF-change, why don't you use a branch (https://www.cvshome.org/docs/manual/cvs-1.12.12/cvs_5.html#SEC56) and merge it to the HEAD after it is ready?

I never know if I can do an upgrade or not. I should use the last stable version, but there are things in CVS that I need, so I cannot use the last stable.

Could you please announce the next compiling CVS-version from CS+CEL before adding new features without fixes or use the branching-mechanism for all these error-prone changes? This would be great, thanks.

[jorrit]

Is there a recent CVS-Version that cleanly compiles? Last time I checked it out, it did compile with errors and my app didn't work because of that.
So I waited one day and checked out again. Ah, that time CS worked fine...but CEL didn't compile anymore because of changes in CS. Then I saw a message that there are errors fixed in CEL depending on the recent changes in CS, so after 5 hours I checked out CEL with these fixes.
But CEL needs a "recent" version from CS so I checked out CS again... and now CS doesn't compile because of further changes...

We're doing heavy changes right now.

Most projects only checkin code that compiles. It doesn't have to work, but it does have to compile at least.
Don't you make a compile-test after changing something? I would never checkin code that doesn't compile, that would keep my team off from work.

Everyone in the CS team checks that it compiles first. The problem is that everyone has different compilers. It is not because something compiles with .NET on windows that it will compile on linux with gcc and vice versa. So that's the main problem.

And when you make severe changes like this SCF-change, why don't you use a branch (https://www.cvshome.org/docs/manual/cvs-1.12.12/cvs_5.html#SEC56) and merge it to the HEAD after it is ready?

Branching in CVS is really not very nice. Also if we do that then the new stuff doesn't get tested as much. i.e. we want as many people as possible to try it out so we can find errors.

I never know if I can do an upgrade or not. I should use the last stable version, but there are things in CVS that I need, so I cannot use the last stable.

Could you please announce the next compiling CVS-version from CS+CEL before adding new features without fixes or use the branching-mechanism for all these error-prone changes? This would be great, thanks.

We will try to stabilize CS as soon as we can.

Greetings,

[morgul]

As of right now, (today) it appears that most of the problems are sorted out. I depend heavily on the CVS of CS and CEL, and I can tell you that I never have to wait more than 2 days before it starts working enough to be used.... and that's not a problem for me. Generally, I can get an error report out, and have someone in the IRC channel fix it within an hour of noticing it's broken. I think that's some amazingly fast turnaround.