Re: [Warzone-dev] How not to use a version control system

2008-11-16 Thread Gerard Krol
Zarel wrote:
> 2008/11/16 Gerard Krol <[EMAIL PROTECTED]>:
>   
>> The untested commit: after a careful reading of the code you conclude
>> that there possibly can't be any bugs left. Unfortunately a trivial bug
>> renders the entire program unusable. Better stay available to do
>> emergency fix-ups.
>>
>> (coloured cursors anyone?)
>>
>> 
>
> Hey, the patch that broke cursors wasn't the colored cursor patch; it
> was a different cursor-hiding patch...
>   
Yeah, by me ;)
I still don't understand why the colored cursor was visible before, as 
it wasn't drawn in the main menu loop. Probably it used the hardware 
cursor in the menu or something like that.

- Gerard


___
Warzone-dev mailing list
Warzone-dev@gna.org
https://mail.gna.org/listinfo/warzone-dev


Re: [Warzone-dev] How not to use a version control system

2008-11-16 Thread Zarel
2008/11/16 Gerard Krol <[EMAIL PROTECTED]>:
> The untested commit: after a careful reading of the code you conclude
> that there possibly can't be any bugs left. Unfortunately a trivial bug
> renders the entire program unusable. Better stay available to do
> emergency fix-ups.
>
> (coloured cursors anyone?)
>

Hey, the patch that broke cursors wasn't the colored cursor patch; it
was a different cursor-hiding patch...

-Zarel

___
Warzone-dev mailing list
Warzone-dev@gna.org
https://mail.gna.org/listinfo/warzone-dev


Re: [Warzone-dev] How not to use a version control system

2008-11-16 Thread Gerard Krol
The untested commit: after a careful reading of the code you conclude 
that there possibly can't be any bugs left. Unfortunately a trivial bug 
renders the entire program unusable. Better stay available to do 
emergency fix-ups.

(coloured cursors anyone?)

Nice list. Now how to make sure people read it

- Gerard

Per Inge Mathisen wrote:
> There are various ways you should not commit things into a version
> control system. Let's go through some of them:
>
> The commit & run: "I have to go now or I'll be late for...", then you
> commit today's work and bolt out the door. A sure way to see to that
> your colleagues sit late or go home early, depending on how close to
> deadline they are.
>
> The blind commit: "I committed *what*?" Always check what changes are
> lurking in your working copy before doing a commit.
>
> The sweet relief commit: After painstakingly long hours of debugging,
> it finally works, and you wrap it up with a commit. It feels good to
> be done. Except you also committed tons of debug code, snarky comments
> and commented out parts, making sure that the next session will be
> even more painstaking. Always read over the whole diff again before
> committing.
>
> The superhuman commit: A truly massive amount of improvements and
> fixes in a single commit, usually signed with a suitably heroic
> one-liner commit message for punch-line. Since it contains so much
> good stuff, nobody has the heart to revert it when they find
> regressions, and due to its massive size, nobody else can bisect it to
> find errors.
>
> The commit flood: Having been bitten by the superhuman commit, you
> split your work instead into dozens if not hundreds of separate
> commits. Guaranteed to make anyone who tries to follow the commit log
> give up in despair, and makes the whole work impossible commit when
> done.
>
> The stroke of genius commit: "Why didn't anyone think of this before?"
> There is usually a good reason. Always sleep on a good idea. It may
> not seem so bright the next morning.
>
> The late night commit: Having *finally* fixed all the bugs and cleaned
> up the code, you write a long and informative commit message, squint
> at the rising sun, and go to bed. Except that your brain is so full of
> diet coke that you have no idea what you were doing, and committed
> from the wrong working copy. Just never commit anything after
> midnight.
>
> The search & replace commit: You find an annoying but unimportant
> frequent mistake in the source code, and fix it by search and replace
> on all occurrances then committing the improvement. As a result, every
> other working copy get conflicts and nothing can be reverted or merged
> past this point.
>
>   - Per
>
> ___
> Warzone-dev mailing list
> Warzone-dev@gna.org
> https://mail.gna.org/listinfo/warzone-dev
>
>   


___
Warzone-dev mailing list
Warzone-dev@gna.org
https://mail.gna.org/listinfo/warzone-dev


Re: [Warzone-dev] How not to use a version control system

2008-11-16 Thread Per Inge Mathisen
On Sun, Nov 16, 2008 at 12:05 PM, Per Inge Mathisen
<[EMAIL PROTECTED]> wrote:
> The commit flood: Having been bitten by the superhuman commit, you
> split your work instead into dozens if not hundreds of separate
> commits. Guaranteed to make anyone who tries to follow the commit log
> give up in despair, and makes the whole work impossible commit when
> done.

Ah, yes. That should really end "impossible to revert when done."

Next up: How not to post rants on an email list...

  - Per

___
Warzone-dev mailing list
Warzone-dev@gna.org
https://mail.gna.org/listinfo/warzone-dev


Re: [Warzone-dev] How not to use a version control system

2008-11-16 Thread Per Inge Mathisen
On Sun, Nov 16, 2008 at 12:14 PM, Kreuvf <[EMAIL PROTECTED]> wrote:
> How about copying it over to the Wiki?

If you find a good place for it, feel free.

  - Per

___
Warzone-dev mailing list
Warzone-dev@gna.org
https://mail.gna.org/listinfo/warzone-dev


Re: [Warzone-dev] How not to use a version control system

2008-11-16 Thread Kreuvf
How about copying it over to the Wiki?

- Kreuvf

___
Warzone-dev mailing list
Warzone-dev@gna.org
https://mail.gna.org/listinfo/warzone-dev