[ITP] bashdb, bash_completion

[Reini Urban]

Christopher Faylor schrieb:
> On Tue, Oct 26, 2004 at 08:34:45PM +0200, Reini Urban wrote:
>>Christopher Faylor schrieb:
>>>I read your email.  I guess my mind boggled at the concept that anyone
>>>would be so..., er, bold as to assume that they could trump someone else's
>>>package by using the name "bash.exe" when there is already an extremely
>>>well-known package which uses that name.
>>>
>>>As I said.  Work this out with Ronald.

I'd love to.

>>>And please don't send any more "ITPs" which wipe out other parts of
>>>people's packages.
>>
>>Why not?
>>Shouldn't it be up to the user to make this decision?
> 
> In theory we try to protect the user from the consequences of bad
> decisions on the part of packaging.

Good point.

> If a user has installed bash and bashdb and then reports a problem with
> bash.exe should you or Ronald get involved?

That's easy.
If it's a bash problem Ronald, if it's a debugger problem me.
This is similar to the practice with the handling of cygwin-specific and 
upstream issues. And I don't see much wrong traffic where users cannot 
decide where to report problems.
Fixes in bash should get into the extended version as well. (Actually 
it's the opposite now. This is the part where I get angry.)

In reality and open-source the upstream maintainer should keep up with 
the pace, or his packages will get replaced by others. Ok, not so with 
bash for sure :) But it happened a decade ago on the shell side also. 
ksh, zsh, even bourne.

> What will cygcheck report if you type "cygcheck -l bash" and bashdb
> is installed?
> 
> What happens when Ronald releases a new version of bash and it wipes
> out the bash.exe that bashdb has just installed?

That's up to the user. The user is always the most intelligent object in 
such decisions. He will complain to Ronald, (debugger, when?), he will 
complain to me (upgrade, when?).

Well, I'm just implementing some missing setup ini options:
Recommends Replaces Build-Depends Maintainer Installed-Size..., but this 
is a technical solution (mostly some Alert) which should be seen 
independently.

> How is Ronald going to feel about your increasing his tech support burden
> by requiring that he spend more time figuring out if a user is talking
> about bash or bashdb?

I don't see any tech support activity (besides you and pierre), really.
How are the users going to feel, that known problems and known fixes are 
not integrated? Not even in experimental test versions? As for example 
Charles or you or Corinna do a really great job with our most important 
packages!

Users are not that stupid and want to try out proposed solutions just to 
check if this might fix their problems or not. And I'm not offering 
solutions, just a new feature, which will most likely not make it into 
the distro or a test release in the next few months, when I read the 
public statements from Roland and you here. Pierre is working for 
himself obviously.

> How much extra mailing list traffic do you think will be generated when
> someone on the cygwin list says "Me am newbie.  Me not find bash.  Me
> need bash." and Larry Hall tells them to use the cygwin package search
> to find bash.exe?

How much traffic is produced now with the current problems with the 
existing package?
How much traffic is produced with not-existing or not-getting upgraded 
packages? coreutils, apache, php, mysql-client and some more.

> (Now, this is the point where everyone scrambles to the the cygwin
> distribution and finds seven instances where a package wipes out another
> package's files and says "See! See! We already DOOOOO IT!!!")

That's the most important point.
Similar to the ploticus and prolog package overwriting /usr/bin/pl.exe
(not yet discussed here to my knowledge)
First serves I would say.
bashdb is a bit different, so I will feel sympathetic if it will
not get into release.
I would feel much better it will get into the official version.

with bash_completion it is easy.
-- 
Reini Urban
http://xarch.tu-graz.ac.at/home/rurban/