----- Original Message -----
From: "Thomas Lawson" <thomas.nigel.lawson at gmail.com>
To: "Dan Tenenbaum" <dtenenba at fredhutch.org>
Cc: "Steffen Neumann" <sneumann at ipb-halle.de>, "bioc-devel" <
Bioc-devel at r-project.org>
Sent: Thursday, April 21, 2016 1:37:15 PM
Subject: Re: [Bioc-devel] linking stage excessive runtime issue on
windows (was: New package submission)
Hi Steffen/Dan
Thank you both for your quick responses and help.
I have just submitted another job and it looks like moscato2 is passing
now. Which is great! However, I am now missing the mac morelia build
completely, which is a bit strange....
- http://staging.bioconductor.org:8000/job/510/
There also seems to be a submission sent at April 21 2016, 11:15 am
(server time) that passed all checks on all systems. I don't think I
submitted this, I am guessing that this was from Dan testing something
Thanks again for your help.
Kind regards,
Tom
On Thu, Apr 21, 2016 at 4:22 PM, Dan Tenenbaum <dtenenba at fredhutch.org>
wrote:
----- Original Message -----
From: "Steffen Neumann" <sneumann at ipb-halle.de>
To: "Thomas Lawson" <thomas.nigel.lawson at gmail.com>, "bioc-devel" <
Bioc-devel at r-project.org>
Sent: Thursday, April 21, 2016 7:29:50 AM
Subject: Re: [Bioc-devel] linking stage excessive runtime issue on
windows (was: New package submission)
Hi Thomas,
mzR has been in stormy waters since the new Windows toolchain
was used. While now all compile errors are gone:
http://bioconductor.org/checkResults/devel/bioc-LATEST/mzR/
The build still timeouts on windows, because for reasons completely
unknown to me the linking step into mzR.dll takes a whopping 20
on windows, and that twice to create the mzR.dll for both i386 and
The package actually INSTALLs in less than 40 minutes (35m23s). And if
look at where it's trying to build, it seems like it's done, and yet
times out. I've seen this before. But also sometimes there are 'real'
timeouts in the build phase, where the build is not complete after 40
minutes.
I don't think this has anything to do with Thomas's issue. I think the
single package builder was installing dependencies without
INSTALL_opts="--merge-multiarch" which I have added and am trying now.
Dan
It is somehow an issue in the mingw compiler. We even tried
an alternative linker (see issue threads linked below).
I don't know if this could be helped with CFLAGS/LDFLAGS.
=> If there is anyone who knows why and if there is any way
to speed this up, any suggestions would be greatly appreciated.
This has been a pain for a long time, and I am sorry
for the turbulences caused.
Yours,
Steffen
We track this issue here:
https://github.com/sneumann/mzR/issues/21#issuecomment-87801932
On Do, 2016-04-21 at 14:37 +0100, Thomas Lawson wrote:
Hi BioConductor community,
I have recently submitted my new R package into BioConductor tracker
(issue
1455) that carries out some data pre-processing for MS/MS spectra.
I seem to be having trouble with the moscato2 Windows build though.
The
package is fine on Linux and Mac but I get an error message (see
below) for
windows. Also, I was not receiving the error last night.
It looks like it has something to do with the BioConductor package
that I
import called "mzR" and an associated DLL it uses.
The error message:
-------------------------------------------
Installing package prior to check...
install for i386
* installing *source* package 'msPurity' ...
** R
** inst
** preparing package for lazy loading
Error in library.dynam(lib, package, package.lib) :
DLL 'mzR' not found: maybe not installed for this architecture?
ERROR: lazy loading failed for package 'msPurity'
* removing
'E:/packagebuilder/workers/jobs/msPurity_20160421061423/msPurity.
buildbin-libdir/msPurity'
Warning: running command 'E:/packagebuilder/R/bin/i386/Rcmd.exe
INSTALL
--library=msPurity.buildbin-libdir msPurity_0.99.0.tar.gz --no-
multiarch'
had status 1
-------------------------------------------
For more information see:
6/
Any help would be really appreciated.
Kind regards,
Tom
[[alternative HTML version deleted]]