Thanks Mike and Herv?!
Somehow, errors in examples that are caused by the state (or absence) of
things on the internet should have a different status in my view than ones
that reflect local state of R and packages - and distinguishing this could
make maintenance of the package corpus & builds more modular.
But I'm not entirely confident whether this view is coherent or
practicable, what do others think?
Wolfgang
2.2.18 12:19, Herv? Pag?s scripsit:
Hi Wolfgang, Mike,
We didn't have a successful build of GenomicFeatures in devel
for many days because of all kinds of problems with the examples
that try to access the Ensembl marts. The latest of which being:
https://support.bioconductor.org/p/105565/
(which also affects GenomicFeatures in release).
Thanks Mike for looking at the timeout issue (FWIW I can reproduce
it). I thought this timeout was a new setting on the Ensembl mart
server side :-)
In GenomicFeatures 1.31.7 I replaced RMySQL with RMariaDB:
https://github.com/Bioconductor/GenomicFeatures/commit/08dd2
4296d94ef31b5f5685240b871c79a160e91
I also made another small speed improvement to makeTxDbFromUCSC().
H.
On 02/02/2018 02:33 AM, Mike Smith wrote:
The error TxDbFromBiomart looks like it might be related to a biomaRt
change I made recently to submit queries using httr rather than RCurl.
Others have reported something similar (e.g
https://urldefense.proofpoint.com/v2/url?u=https-3A__support
.bioconductor.org_p_104502_&d=DwIGaQ&c=eRAMFD45gAfqt84VtBcfh
Q&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=nTaIFim9PZ
DsGaU0xtK3D4cw4UKBrPzen-jCDV-ppDE&s=k3wnoKsf0v7Biv9DKASzF3h
SAzhO02fR9UMIiznM2EA&e=) and I raised the timeout from
10 to 60 seconds. I guess with the old version it was even longer than
that.
I haven't been able to recreate the problem at my end, I think the time
taken is related as much to the internet connection as to the query, but
I'll take a look at the failing example to see if I can shed any more
light
on it.
Mike
On 2 February 2018 at 10:41, Wolfgang Huber <wolfgang.huber at embl.de>
wrote:
Thanks Herv?!
This seems to take a long time to propagate. As of now,
https://urldefense.proofpoint.com/v2/url?u=https-3A__biocond
uctor.org_packages_devel_bioc_html_GenomicFeatures.html&d=DwIGaQ&c=
eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJ
KaaPhzWA&m=nTaIFim9PZDsGaU0xtK3D4cw4UKBrPzen-jCDV-ppDE&s=
09LGf-6SKIrEdIR2hwC4Vq1YOBj-3PrsuCanwxv9AMs&e=
still shows 1.31.3.
( Btw, there's also an error in the build report which seems to come
from
a rather excessive-looking example in the makeTxDbFromBiomart man page -
that maybe could better live in a vignette, not least for reducing
brittleness? )
Best wishes
Wolfgang
30.1.18 19:00, Herv? Pag?s scripsit:
This is done in GenomicFeatures 1.31.6.
Note that I also made a few changes to makeTxDbFromUCSC() to make it
a little bit faster (about 2x).
@Kasper: moving the makeTxDb* functions to a GenomicFeaturesBuildTools
or GenomicFeaturesForge package is maybe an idea to explore...
H.
On 01/26/2018 06:09 PM, Kasper Daniel Hansen wrote:
As an alternative to Suggests, perhaps make a
GenomicFeaturesBuildTools.
Not sure if it is better or worse, just different
On Fri, Jan 26, 2018 at 2:39 PM, Wolfgang Huber <
wolfgang.huber at embl.de>
wrote:
26.1.18 14:59, Martin Morgan scripsit:
On 01/24/2018 03:38 PM, Wolfgang Huber wrote:
GenomicFeatures_1.31.3 imports RMySQL.
I'm having great trouble installing RMySQL from source on a recent
MacOS
(10.13.3) with homebrew.
The package's homepage says "The 'RMySQL' package contains an old
implementation based on legacy code from S-PLUS which being phased
out. A
modern 'MySQL' client based on 'Rcpp' is available from the
'RMariaDB'
package" https://urldefense.proofpoint.
com/v2/url?u=https-3A__cran.r-
2Dproject.org_web_packages_RMySQL_index.html&d=DwICAg&c=eRAM
FD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaP
hzWA&m=jUD3XGyxapAK8nfWvxbR_9Rz7RBxP592NO0nWz_wyr0&s=0neTB2B
Jghm4wF-LnEZ9V9sV_AbF1botcOeTb_bWk-w&e=
So is it time to heed that advice and migrate GenomicFeatures to
RMariaDB ?
Out of curiosity, is MariaDB easier to install on your system? Its
system
dependencies are described at https://urldefense.proofpoint.
com/v2/url?u=https-3A__urldefense.proofpoint&d=DwIGaQ&c=eRAM
FD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaP
hzWA&m=nTaIFim9PZDsGaU0xtK3D4cw4UKBrPzen-jCDV-ppDE&s=t1JSruT
yZxTGystW_bcaK5XfEnAVD0dTfxdLOO3boT4&e=.
com/v2/url?u=https-3A__CRAN.R-2Dproject.org_package-3DRMaria
DB&d=DwICAg&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJ
YbW0WYiZvSXAJJKaaPhzWA&m=jUD3XGyxapAK8nfWvxbR_9Rz7RBxP592NO0
nWz_wyr0&s=55tLfrRpaigMug-pW7L1AhpsD1A830xw1f9d1FZEANo&e=
I have no problems installing RMariaDB on MacOS (10.13.3) on an
R-devel
from source, after installing mariadb-connector-c with homebrew.
OTOH, I have not figured out a way to install RMySQL neither on my
R-devel
from source (various complaints about missing .h files) nor on a
binary
R
3.4.2 with the binary package download (complaints about missing
system
libraries / wrong versions).
Thanks and kind regards
Wolfgang
FWIW MySQL is a relatively recent addition as a dependency to
GenomicFeatures; it enables `makeTxDbFromEnsembl()`, which is
probably
a
much more stable solution than `makeTxDbFromBiomart()`. On the other
had
Johannes does an excellent job on the ensembldb packages, so perhaps
this
code could really be conditional with the RMySQL dependency moved to
Suggests:
Martin
With thanks in advance-
Wolfgang
-------
Wolfgang Huber
Principal Investigator, EMBL Senior Scientist
European Molecular Biology Laboratory (EMBL)
Heidelberg, Germany
wolfgang.huber at embl.de
https://urldefense.proofpoint.com/v2/url?u=http-3A__www.hube
r.embl.de&d=DwICAg&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWd
GbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=jUD3XGyxapAK8nfWvxbR_9Rz7RBx
P592NO0nWz_wyr0&s=Vp0j7sKJXiZSWHn3LVWSECRS-f-AFZZSSh0mq5rvJ-0&e=