[U-Boot] [PATCH 2/2] beagle: handle import of environments in files with CRLF as line endings

Wolfgang Denk wd at denx.de
Sun May 13 23:38:11 CEST 2012


Dear Alexander,

In message <4FB01720.90402 at ahsoftware.de> you wrote:
>
> Sorry, it seems you just are unable or yiu don't want to understand the 
> problem. So here is my last message on this topic, trying to explain it 
> for you.

I'm sorry you are giving up so early, just one resubmit before getting
this into an acceptable state.

> So one just tells the user(s) that if they want to change some option 
> they just have to create a text file on a specific place which contains 
> a line like e.g.

I guess this is the point where the problem starts.  You should not
write anywhere "they just have to create a text file", because "a text
file" is not a precise enough definition of the required input format.

> The only problem is that such a description currently only works for 
> Linux users. ...

The problem is not Windows users versus Linux users here.  The problem
is unsufficient information.

>          ... I don't know with what devices and users you are dealing, 
> but there are some people out in the wild which don't know (much) about 
> Linux, or even about the difference in line endings between text files 
> creates using Windows or Linux. Some of them even just want to use their 
> device.

In either case, people are usually pretty good in following pre-canned
recipies for doing things.

Instead of "create a text file" one could for example document that "a
UNIX style text file, i. e. with only "\n" line endings (and not the
DOS-derived "\r\n" line endings) is needed".  One could add some
description that "on Windows systems, the command ... can be used to
convert a DOS file into this format" (and provide a URL where to get
this tool), "while on UNIX systems like Linux the dos2unix tool can be
used".

If you then add an example suitable for copy & paste you can solve a
very large percentage of the propblems you see now.

> And in addition, the resulting problems hare very hard to diagnose, 
> because there will be no obvious error message when something contains a 
> trailing CR.

What exactly are such error modes, by the way?  I would expect that
trailing white space is pretty much harmless for most variable
settings?

> Now I needed about 10 mails to describe a problem from which I thought 
> it's totally obvious. Sorry, but I accept that I'm totally unable to 
> deal with this list in any, for me reasonable, time frame.

What you provide is valuable input, and it really makes sense to
discuss such issues.  I would like to bring such discussions to a
point where either we understand why your approach makes sense, or
where you agree that there are alternatives that solve the problem as
well (or better) as your original proposal.  However, this is
difficult to acchieve if you give up early.  And believe me, this is
taking my time (and Marek's) as much as yours.  We do this not to
annoy you, but because we are interested in the best possible
solution.

Best regards,

Wolfgang Denk

-- 
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
There's an old story about the person who wished his computer were as
easy to use as his telephone. That wish has come  true,  since  I  no
longer know how to use my telephone.


More information about the U-Boot mailing list