Reinhard | 1 Apr 2004 01:07
Picon
Favicon

Re: Wether to apply changes from newxml to official profile??

Thanks for your attention.

On Wednesday 31 March 2004 11:42, Jamie Bennett wrote:
> Reinhard wrote on 31 March 2004 09:47
>
> > Get a Makefile from the book, is really trivial, compared with
> > the command translation. But I'm on the way - and only few commands
> > pending for workout.
>
> All though not an ideal solution, have you looked at Vasilli's YAALFS
> implementation? (http://vassilidzuba.nerim.net/yaalfs.html) It could
> convert nALFS profiles into bash commands. You could probaby reverse it?

Noop - cause IMHO the xml-side is always well structured and in order, so it's 
quite easy to transform xml to anything you want.

Shellscripts are not at all well structured and they are seldom ordered. Quite 
often tricky variants are used, so if you want to use the full bandwidth of 
alfs profile syntax, you have to check for most system calls, whether you 
could write it in alfs-syntax or whether you have to (fallback to)  use 
*execute*.

Just an example (not sure, whether from lfs or blfs):
I dumped all sed-usages to a file to investigate, what could be written with 
*search_replace* and from over 30 sed-variants, only one could be written 
with *search_replace* - well, without human intervention and that is my 
focus. 
If I have to edit a single file from the generated profiles (except the config 
files), I would consider my work as useless.

(Continue reading)

James Robertson | 2 Apr 2004 01:02
Picon
Favicon

Re: [VOTE] Re: directory-handling with nALFS

Kevin P. Fleming wrote:
+1 for leave it alone.  I like the error and can easily use the parents 
option

James
--

-- 
http://linuxfromscratch.org/mailman/listinfo/alfs-discuss
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Reinhard | 2 Apr 2004 07:03
Picon
Favicon

Re: [VOTE] Re: directory-handling with nALFS

On Friday 02 April 2004 01:02, James Robertson wrote:
> +1 for leave it alone.  I like the error and can easily use the parents
> option
>

Hello James,

could you please explain me the benefit of the error?

The change I did was to suppress the error only in the case that the existing 
target already is a directory. If the target exists, but is not a directory, 
an error will stil occure.

I suppose, that you do your own profile-writing.
Do you use the *parents* option for every mkdir occurrence?
I'm working to create profiles from the book and for so, I did not want to 
change the content of the book (adding a *parents* option where it is not in 
the book, is changing the book-content - IMHO).

What do you think about an entry in nALFSrc, whether the profile-handling 
should be "strict" or "relaxed" - in the sense, that an "relaxed" operation 
is like the behaviour of a well-formed Makefile :)

Don't know, whether I'm allowed to vote.

If so, I vote for relaxed handling (the patch and if liked the nALFSrc 
extension).

The fix I did was a first step toward this, the entry in nALFSrc could be a 
final solution for relaxed handling.
(Continue reading)

Reinhard | 2 Apr 2004 07:18
Picon
Favicon

questions about nALFS

On Wednesday 31 March 2004 15:32, Kevin P. Fleming wrote:
> (We do have a Bugzilla system for this type of thing as well Reinhard,
> you should probably check that out :-))

I'll check it out.

Kevin, is your "todo"-list or roadmap publicly available?

I've lot of questions related to nALFS ...
only few here:

- exists something like a screenplay for nALFS?

- the current handler for search_replace uses "selfwritten" replaces. This 
handler could not search for patterns. How about to change it to use the 
system's sed?

- also I like to repeat my question about filename-handling for search/
replace:
Don't know, why most occurences of sed don't use the -i option but copy the 
file away and use different files for in and out (may be cause of 
sideeffects) - but sometimes sed is used with 2 really different files. One 
to read from and the other to write to.
What do you think about the possibility to be able to specify such in an nALFS 
profile? (i.e. something like *infile* and *outfile*)

tbc...

Kind regards

(Continue reading)

Reinhard | 2 Apr 2004 08:02
Picon
Favicon

Re: Bugzilla link

On Friday 02 April 2004 07:18, Reinhard wrote:
> On Wednesday 31 March 2004 15:32, Kevin P. Fleming wrote:
> > (We do have a Bugzilla system for this type of thing as well Reinhard,
> > you should probably check that out :-))

May be I need some new glasses, but I could not find any link to BZ.
Could somebody please send me such a link?

Thanks in advance!

Kind regards

Reinhard

--

-- 
http://linuxfromscratch.org/mailman/listinfo/alfs-discuss
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Wingman | 2 Apr 2004 10:58

space for building LFS

Hello,

I've just built a LFS with the CVS version 2004-03-03. Works ok for me. It
was built using a /mnt/lfs 5GB partition.
When trying to use a 1,5GB partition it broke on glibc chapter 6.

How much space do LFS needs ? I thought it was about 1GB..
Is there a difference (in space use) between using the book or nALFS ?

Thx.

--

-- 
http://linuxfromscratch.org/mailman/listinfo/alfs-discuss
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Reinhard | 2 Apr 2004 13:10
Picon
Favicon

formatting in syntax_doc

Hello,

last time I worked a bit with the syntax doc (HTML-version).
In toc the tags play little role. The 'Element' is the eyecatcher cause of the 
indentation and its optical weight.

So what do you think about emphasizing the tags? 

Within all the links it's now a pain for the eyes to find a tag - nearly 
impossible without the textsearch of the browser.

Kind regards 

Reinhard

--

-- 
http://linuxfromscratch.org/mailman/listinfo/alfs-discuss
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Kevin P. Fleming | 2 Apr 2004 15:35
Picon
Favicon

Re: Bugzilla link

Reinhard wrote:

> May be I need some new glasses, but I could not find any link to BZ.
> Could somebody please send me such a link?
> 

It just recently got added to the website pages for ALFS, and still 
doesn't appear on the main page for some reason (I have reported that).

In any case, all LFS Bugzilla stuff is available at

http://bugs.linuxfromscratch.org
--

-- 
http://linuxfromscratch.org/mailman/listinfo/alfs-discuss
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Kevin P. Fleming | 2 Apr 2004 15:36
Picon
Favicon

Re: space for building LFS

Wingman wrote:

> How much space do LFS needs ? I thought it was about 1GB..
> Is there a difference (in space use) between using the book or nALFS ?

It shouldn't matter whether you follow the book manually or use an ALFS 
profile, the same steps are being followed. If this used more space than 
the book says it will, you should report that to lfs-dev so the book can 
be updated.
--

-- 
http://linuxfromscratch.org/mailman/listinfo/alfs-discuss
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Kevin P. Fleming | 2 Apr 2004 15:40
Picon
Favicon

Re: questions about nALFS

Reinhard wrote:

> Kevin, is your "todo"-list or roadmap publicly available?

The "todo" list is primarily in Bugzilla, at least for all the things 
that will happen in the near future.

> - exists something like a screenplay for nALFS?

I don't know what this means :-)

> - the current handler for search_replace uses "selfwritten" replaces. This 
> handler could not search for patterns. How about to change it to use the 
> system's sed?

This has been a topic of discussion recently; I don't know whether it 
would be better to use sed for this, or just build nALFS against the 
regex or PCRE libraries and do the search-replace with regexes in nALFS 
itself.

> - also I like to repeat my question about filename-handling for search/
> replace:
> Don't know, why most occurences of sed don't use the -i option but copy the 
> file away and use different files for in and out (may be cause of 
> sideeffects) - but sometimes sed is used with 2 really different files. One 
> to read from and the other to write to.

I think this is probably a good idea, making the output file path 
optional would work. Please enter an enhancement request in Bugzilla so 
we won't forget about it.
(Continue reading)


Gmane