[pkg-bioc] CVS %p

[Steffen Moeller]

On Monday 12 February 2007 01:42:20 Dirk Eddelbuettel wrote:
> (CC back to list added)
>
> On 12 February 2007 at 01:25, Steffen Moeller wrote:
> | I had indeed read through that help file. Your recollection of wget
> | behaviour is in sync with mine, I am afraid.  --mirror should make it
> | stop that.
> AFAICR --mirror does weird things placing everything several subdirs down.
Hm. Not if it is passed the right arguments.  "man wget" helps us Schlaumeiers 
(a word that should urgently make it into the English Sprachschatz (another  
word that should (less) urgently make it into the English Sprachschatz)), the 
option to go for is "-nc" for "no clobber":

moeller at pc02:~$ wget -nc 
http://bioconductor.org/packages/1.9/bioc/src/contrib/biocViews_1.2.0.tar.gz
--09:54:15--  
http://bioconductor.org/packages/1.9/bioc/src/contrib/biocViews_1.2.0.tar.gz
Resolving bioconductor.org... 140.107.3.20
Connecting to bioconductor.org|140.107.3.20|:80... connected.
HTTP request sent, awaiting response... 200 OK
File `biocViews_1.2.0.tar.gz' already there; not retrieving.

Aborted (core dumped)

The fun bit is that is throws this core. It is acting right, though. This is 
what strace says:

write(2, "File `biocViews_1.2.0.tar.gz\' al"..., 62File 
`biocViews_1.2.0.tar.gz' already there; not retrieving.

) = 62
time(NULL)                              = 1171270465
stat64("/etc/localtime", {st_mode=S_IFREG|0644, st_size=837, ...}) = 0
rt_sigprocmask(SIG_UNBLOCK, [ABRT], NULL, 8) = 0
tgkill(7023, 7023, SIGABRT)             = 0
--- SIGABRT (Aborted) @ 0 (0) ---
+++ killed by SIGABRT (core dumped) +++
Process 7023 detached

> Anyway, if and when we rewrite this from R we can always test with
> file.exists()
>
> | Hm. But you can just delete the build directory to which the orig.tar.gz
> | are only links, not the real files. Thus: it is all fine just to rerun
> | the script after a "rm -r builds", not need to redownload bits.
>
> Somehow a restart earlier lead to double downloads.
I don't get it. This may be true for CRAN, but not for BioC. The downloading 
is only performed by the R script. When running "perl cran2deb.pl", this is 
not executed. Am I missing something?

What _is_ performed twice in BioC though is the compilation of the packages. 
This I consider unfortunate. Somehow the --ignore flag seems to be 
influencing the behaviour since it is fine when specifying subsets 
with --subset (which I started to do in routine now). 

> On the other hand the 
> outer loop over package builds attempts seems misguiding: It is starting
> package with external depends like r-bioc-biobase which I haven't even
> built yet ...
We (c|sh)ould test for the installaion-status of any package of the external 
depends, though.  This is about what I meant in my previous emails.

> | Did you manage to get the descriptions of bioc back to shape? This is
> | rather embarrassing. Next week.
>
> I don't follow. But next week sounds good. Get some sleep!!

When you have a look at http://pc02.inb.uni-luebeck.de:8080/pkg-bioc/Packages 
you will find the descriptions of the BioC packages to be rather 
non-existing. The CRAN are fine.

Cheers,

Steffen

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://lists.alioth.debian.org/pipermail/pkg-bioc-devel/attachments/20070212/a889280c/attachment.pgp