Re: [Bioc-devel] Unit tests pass locally but fail on Bioconductor machines

[Christian Holland]

Dear Vincent,

you've hit the nail on the head! Indeed the error was raised by the fact that 
ExpressionSet objects store the R version. I followed your suggestion and 
create this object now every time I run the tests.

And also thanks @Lori, part of the solution was indeed to update to R4.0

Thanks,
Christian

> On 7. Apr 2020, at 14:17, Vincent Carey  wrote:
> 
> I will chime in because the error is a bit obscure.  You serialized 
> ExpressionSet instances for testing.  In the expected_res
> in your tests we have
> 
> 
>  ..@ .__classVersion__:Formal class 'Versions' [package "Biobase"] with 1 slot
>   .. .. ..@ .Data:List of 4
>   .. .. .. ..$ : int [1:3] 3 6 1
>   .. .. .. ..$ : int [1:3] 2 46 0
>   .. .. .. ..$ : int [1:3] 1 3 0
>   .. .. .. ..$ : int [1:3] 1 0 0
> 
> which formally encodes the version of R with which the ExpressionSet was 
> created.  The one created in the test on the
> build system will have first list element c(4,0,0)
> 
> How to get around this for the long run?  One possibility is to create your 
> S4 instances from serializations of more basic objects as 
> part of the test process.
> 
> On Tue, Apr 7, 2020 at 7:55 AM Shepherd, Lori  > wrote:
> There are many changes to base R hence the switch from 3.6 to 4.0.   I would 
> highly suggest using R 4.0.  There is a R-alpha version now available for all 
> platforms with the R - 4.0 release schedule for later this month.  Please use 
> the R-alpha version .
> 
> on CRAN.r-project.org  you should see a section
> Sources of R alpha and beta releases (daily snapshots, created only in time 
> periods before a planned release) with links to R-alpha for linux.  There is 
> a link here to R-alpha for mac  http://mac.r-project.org/ 
>    and one here for windows:  
> https://cran.rstudio.com/bin/windows/base/rpatched.html 
> 
> 
> This should allow you to download the most stable R-4.0 alpha before the 
> release at the end of the month.
> Hope this helps.
> 
> 
> 
> 
> Lori Shepherd
> 
> Bioconductor Core Team
> 
> Roswell Park Comprehensive Cancer Center
> 
> Department of Biostatistics & Bioinformatics
> 
> Elm & Carlton Streets
> 
> Buffalo, New York 14263
> 
> 
> From: Bioc-devel  > on behalf of Christian Holland 
> mailto:cholland2...@gmail.com>>
> Sent: Tuesday, April 7, 2020 7:22 AM
> To: bioc-devel@r-project.org  
> mailto:bioc-devel@r-project.org>>
> Subject: [Bioc-devel] Unit tests pass locally but fail on Bioconductor 
> machines
> 
> Hi there,
> 
> the unit tests (implemented with testthat) of my package 
> (https://github.com/saezlab/dorothea  
> >) 
> run smoothly on local machines (tested for Linux, Windows and macOS). 
> However, on the Bioconductor machines (both Linux and Window) a particular 
> test fails. The feature that distinguish this particular test from all other 
> (that passed without any problems) is the use of the ExpressionSet class from 
> the Biobase package. Locally I am using the Biobase version 2.46.0.
> 
> You can see the build output from Bioconductor here: 
> http://bioconductor.org/spb_reports/dorothea_buildreport_20200407063358.html 
> 
>  
>  >
> 
> Is it possible that this error is somehow related to the used R Version? 
> Locally I am still using 3.6.2 but if I understand correctly the package is 
> build and checked on Bioconductor machines with R version 4.0. So far I 
> refrained from installing the development version of R4.0 as it crashed 
> RStudio for a colleague of mine.
> 
> Many thanks for your support.
> 
> Christian
> 
> 
> 
> [[alternative HTML version deleted]]
> 
> ___
> Bioc-devel@r-project.org  mailing list
> https://stat.ethz.ch/mailman/listinfo/bioc-devel 
> 
> 
> 
> This email message may contain legally privileged and/or confidential 
> information.  If you are not the intended recipient(s), or the employee or 
> agent responsible for the delivery of this message to the intended 
> recipient(s), you are hereby notified that any disclosure, copying, 
> distribution, or use of this email message is prohibited.  If you have 
> received this message in error, please notify the sender immediately by 
> e-mail and delete this email message from your computer. Thank you.
> [[alternative HTML version deleted]]
> 
> 

Re: [Bioc-devel] Unit tests pass locally but fail on Bioconductor machines

[Vincent Carey]

I will chime in because the error is a bit obscure.  You serialized
ExpressionSet instances for testing.  In the expected_res
in your tests we have


 ..@ .__classVersion__:Formal class 'Versions' [package "Biobase"] with 1
slot

  .. .. ..@ .Data:List of 4

  .. .. .. ..$ : int [1:3] 3 6 1

  .. .. .. ..$ : int [1:3] 2 46 0

  .. .. .. ..$ : int [1:3] 1 3 0

  .. .. .. ..$ : int [1:3] 1 0 0


which formally encodes the version of R with which the ExpressionSet was

created.  The one created in the test on the

build system will have first list element c(4,0,0)


How to get around this for the long run?  One possibility is to create your

S4 instances from serializations of more basic objects as

part of the test process.

On Tue, Apr 7, 2020 at 7:55 AM Shepherd, Lori 
wrote:

> There are many changes to base R hence the switch from 3.6 to 4.0.   I
> would highly suggest using R 4.0.  There is a R-alpha version now available
> for all platforms with the R - 4.0 release schedule for later this month.
> Please use the R-alpha version .
>
> on CRAN.r-project.org you should see a section
> Sources of R alpha and beta releases (daily snapshots, created only in
> time periods before a planned release) with links to R-alpha for linux.
> There is a link here to R-alpha for mac  http://mac.r-project.org/   and
> one here for windows:
> https://cran.rstudio.com/bin/windows/base/rpatched.html
>
> This should allow you to download the most stable R-4.0 alpha before the
> release at the end of the month.
> Hope this helps.
>
>
>
>
> Lori Shepherd
>
> Bioconductor Core Team
>
> Roswell Park Comprehensive Cancer Center
>
> Department of Biostatistics & Bioinformatics
>
> Elm & Carlton Streets
>
> Buffalo, New York 14263
>
> 
> From: Bioc-devel  on behalf of
> Christian Holland 
> Sent: Tuesday, April 7, 2020 7:22 AM
> To: bioc-devel@r-project.org 
> Subject: [Bioc-devel] Unit tests pass locally but fail on Bioconductor
> machines
>
> Hi there,
>
> the unit tests (implemented with testthat) of my package (
> https://github.com/saezlab/dorothea )
> run smoothly on local machines (tested for Linux, Windows and macOS).
> However, on the Bioconductor machines (both Linux and Window) a particular
> test fails. The feature that distinguish this particular test from all
> other (that passed without any problems) is the use of the ExpressionSet
> class from the Biobase package. Locally I am using the Biobase version
> 2.46.0.
>
> You can see the build output from Bioconductor here:
> http://bioconductor.org/spb_reports/dorothea_buildreport_20200407063358.html
> <
> http://bioconductor.org/spb_reports/dorothea_buildreport_20200407063358.html
> >
>
> Is it possible that this error is somehow related to the used R Version?
> Locally I am still using 3.6.2 but if I understand correctly the package is
> build and checked on Bioconductor machines with R version 4.0. So far I
> refrained from installing the development version of R4.0 as it crashed
> RStudio for a colleague of mine.
>
> Many thanks for your support.
>
> Christian
>
>
>
> [[alternative HTML version deleted]]
>
> ___
> Bioc-devel@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/bioc-devel
>
>
> This email message may contain legally privileged and/or confidential
> information.  If you are not the intended recipient(s), or the employee or
> agent responsible for the delivery of this message to the intended
> recipient(s), you are hereby notified that any disclosure, copying,
> distribution, or use of this email message is prohibited.  If you have
> received this message in error, please notify the sender immediately by
> e-mail and delete this email message from your computer. Thank you.
> [[alternative HTML version deleted]]
>
> ___
> Bioc-devel@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/bioc-devel
>

-- 
The information in this e-mail is intended only for the ...{{dropped:18}}

___
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel

Re: [Bioc-devel] Unit tests pass locally but fail on Bioconductor machines

[Shepherd, Lori]

There are many changes to base R hence the switch from 3.6 to 4.0.   I would 
highly suggest using R 4.0.  There is a R-alpha version now available for all 
platforms with the R - 4.0 release schedule for later this month.  Please use 
the R-alpha version .

on CRAN.r-project.org you should see a section
Sources of R alpha and beta releases (daily snapshots, created only in time 
periods before a planned release) with links to R-alpha for linux.  There is a 
link here to R-alpha for mac  http://mac.r-project.org/   and one here for 
windows:  https://cran.rstudio.com/bin/windows/base/rpatched.html

This should allow you to download the most stable R-4.0 alpha before the 
release at the end of the month.
Hope this helps.




Lori Shepherd

Bioconductor Core Team

Roswell Park Comprehensive Cancer Center

Department of Biostatistics & Bioinformatics

Elm & Carlton Streets

Buffalo, New York 14263


From: Bioc-devel  on behalf of Christian 
Holland 
Sent: Tuesday, April 7, 2020 7:22 AM
To: bioc-devel@r-project.org 
Subject: [Bioc-devel] Unit tests pass locally but fail on Bioconductor machines

Hi there,

the unit tests (implemented with testthat) of my package 
(https://github.com/saezlab/dorothea ) run 
smoothly on local machines (tested for Linux, Windows and macOS). However, on 
the Bioconductor machines (both Linux and Window) a particular test fails. The 
feature that distinguish this particular test from all other (that passed 
without any problems) is the use of the ExpressionSet class from the Biobase 
package. Locally I am using the Biobase version 2.46.0.

You can see the build output from Bioconductor here: 
http://bioconductor.org/spb_reports/dorothea_buildreport_20200407063358.html 


Is it possible that this error is somehow related to the used R Version? 
Locally I am still using 3.6.2 but if I understand correctly the package is 
build and checked on Bioconductor machines with R version 4.0. So far I 
refrained from installing the development version of R4.0 as it crashed RStudio 
for a colleague of mine.

Many thanks for your support.

Christian



[[alternative HTML version deleted]]

___
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel


This email message may contain legally privileged and/or confidential 
information.  If you are not the intended recipient(s), or the employee or 
agent responsible for the delivery of this message to the intended 
recipient(s), you are hereby notified that any disclosure, copying, 
distribution, or use of this email message is prohibited.  If you have received 
this message in error, please notify the sender immediately by e-mail and 
delete this email message from your computer. Thank you.
[[alternative HTML version deleted]]

___
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel

Re: [Bioc-devel] Unit Tests & Test Coverage

[Romero, Juan Pablo]

Hi Hervé,


Thanks for all the clarifications related to the tests and the coverage for 
packages.


Yes, I've just checked EventPointer landing page and the coverage bad is 
working great at

this moment. It seems that it was just a matter of time before it was displayed.


Again, thanks for all.


Best regards,


Juan Pablo


From: Hervé Pagès <hpa...@fredhutch.org>
Sent: Friday, March 24, 2017 8:29:08 AM
To: Romero, Juan Pablo; bioc-devel@r-project.org
Subject: Re: [Bioc-devel] Unit Tests & Test Coverage

The coverage badge finally got updated:

   https://bioconductor.org/packages/3.5/bioc/html/EventPointer.html

I didn't do anything. Looks like for some reason it took a long time
for the coverage result on codecov.io to propagate to the package
landing page.

Cheers,
H.


On 03/23/2017 01:32 PM, Hervé Pagès wrote:
> Hi Juan Pablo,
>
> Thanks for adding unit tests to your package. We really encourage
> developers to do this. I'm sorry that this is not reflected on
> EventPointer's landing page.
>
> I investigated this a bit (I'm not familiar with how the coverage
> badges are generated on the package landing pages) and here is
> what I found so far.
>
> One possible reason for this is that some packages on
> Bioconductor-mirror are lagging behind their latest svn revision.
> This is actually the case right now for EventPointer. Latest svn
> commit is 127626:
>
>   https://hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/EventPointer
>
> but the package is still at commit 127616 on Bioconductor-mirror:
>
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_Bioconductor-2Dmirror_EventPointer=DwIGaQ=eRAMFD45gAfqt84VtBcfhQ=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA=QY4w9Io4Cx1vbUkb-sjYFCC4fp88dTA9tPfm7O_ueSQ=O1j0kLptbtmlzKItTODgtPNL9X7OXDteiR7Sn7zCgMg=
>
> This is a known issue. We'll try to remedy this ASAP.
>
> This can be a reason why coverage is not updated on a package
> landing page. That's because the script we use for computing
> the coverage interacts with our svn repo (on hedgehog.fhcrc.org)
> in order to get the latest revision of the package, but also with
> the Bioconductor-mirror account on github
> (https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_Bioconductor-2Dmirror_=DwIGaQ=eRAMFD45gAfqt84VtBcfhQ=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA=QY4w9Io4Cx1vbUkb-sjYFCC4fp88dTA9tPfm7O_ueSQ=7KRBTOWvXmA76VMkUfMzdWDh-_KsXNXV8K2qWDirx6o=
> ) in order to map the
> latest svn revision of a package to the corresponding git commit
> on Bioconductor-mirror. This seems to be needed in order to upload
> the coverage results to codecov.io where the coverage history is
> kept e.g.
>
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__codecov.io_github_Bioconductor-2Dmirror_VariantAnnotation_branch_master=DwIGaQ=eRAMFD45gAfqt84VtBcfhQ=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA=QY4w9Io4Cx1vbUkb-sjYFCC4fp88dTA9tPfm7O_ueSQ=ANws4fvdvik6ga1hynXgSPss0Jrix_9wdl5_tl_0IIo=
>
> The coverage information displayed on a package landing page is coming
> from there.
>
> When a package is lagging behind on Bioconductor-mirror, the mapping
> between last svn revision and git commit fails and so uploading the
> coverage results to codecov.io also fails.
>
> However, in the case of EventPointer, there seems to be already 1 point
> in the coverage history at codecov.io:
>
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__codecov.io_github_Bioconductor-2Dmirror_EventPointer_branch_master=DwIGaQ=eRAMFD45gAfqt84VtBcfhQ=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA=QY4w9Io4Cx1vbUkb-sjYFCC4fp88dTA9tPfm7O_ueSQ=dDjRgkJSvMLudzs009l-1rcMyEWa3psee7CK90enRk0=
>
> It shows a coverage of about 30%. For some reason, this doesn't show
> up on EventPointer's landing page. I'll investigate more and will
> let you know.
>
> H.
>
>
> On 03/23/2017 04:58 AM, Romero, Juan Pablo wrote:
>> Hi,
>>
>>
>> I'm currently doing the last updates to my package EventPointer before
>> the deadline for next BioC release.
>>
>>
>> I've added unit tests according to the guide lines in
>> https://urldefense.proofpoint.com/v2/url?u=http-3A__bioconductor.org_developers_unitTesting-2Dguidelines_=DwIFaQ=eRAMFD45gAfqt84VtBcfhQ=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA=4xEe7s13qw8OtBMzGMqJ0vk6PIoU68FR0RJ88otY0yg=47DUypCrCnkcv9C1Mlybb0QHi3RLMZh57Q35XoIXkBU=
>> and the package
>>
>> build report shows no problems for both build and check in all
>> systems. However, the test coverage tab at top of the package page
>>
>> still displays "unknown" and im not sure if there is any problem with
>> the tests or how to find possible errors.
>>
>>
>> Here is what the build report shows in all O

Re: [Bioc-devel] Unit tests

[Kasper Daniel Hansen]

There are multiple examples of unit tests for parsers.  In your case I
would include a small BAM file in /inst/extdata and make a test which
parses this file and asserts a number of things about the output.
Depending on philosophy, this may not be considered a pure unit test, but
it is very useful.  If the BAM file gets too big if its going to be
"interesting" we have in the past released experiment data packages with
testing data; this has usually been done for packages which read multiple
different files.  Examples are IlluminaDataTestFiles and
AffymetrixDataTestFiles (for parsing microarray data)

On Thu, Mar 23, 2017 at 10:37 PM, Dan Tenenbaum 
wrote:

> Here's what BiocCheck says:
>
> Consider adding unit tests. We strongly encourage them. See
> http://bioconductor.org/developers/how-to/unitTesting-guidelines/
>
> So, unit tests are encouraged but not required.
> Dan
>
>
> - Original Message -
> > From: "张腾" <2573552...@qq.com>
> > To: "bioc-devel" 
> > Sent: Thursday, March 23, 2017 7:33:22 PM
> > Subject: [Bioc-devel] Unit tests
>
> > Hello all,
> >
> > I am not sure whether the Unit tests are necessary to release a package
> for
> > Bioconductor. In my package, only one main function to export and other
> > functions are hidden into the main function. In my package, the input
> data are
> > some files in BAM format and the output is a report in HTML format. So I
> want
> > to know if the unit tests are very important and necessary for releasing
> a
> > package. If it is necessary, how can I add these tests to my package in
> this
> > special input and output situation?
> >
> > Can anyone help me solve this problem? Thank you very much
> >   [[alternative HTML version deleted]]
> >
> > ___
> > Bioc-devel@r-project.org mailing list
> > https://urldefense.proofpoint.com/v2/url?u=https-3A__stat.
> ethz.ch_mailman_listinfo_bioc-2Ddevel=DwICAg=eRAMFD45gAfqt84VtBcfhQ=
> TF6f93hjWmgMzjqP9F3thRifibmFvfjc5Ae-bzNwDGo=-
> y21Sirq2bdhul9Ar8vnDiy_LjASe_ND5ozxWVwudWw=Wroy1VJao_
> ttDgKZZY7vOApZkTTauP_S8kAgYTs6jo4=
>
> ___
> Bioc-devel@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/bioc-devel

[[alternative HTML version deleted]]

___
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel

Re: [Bioc-devel] Unit tests

[Dan Tenenbaum]

Here's what BiocCheck says:

Consider adding unit tests. We strongly encourage them. See
http://bioconductor.org/developers/how-to/unitTesting-guidelines/

So, unit tests are encouraged but not required.
Dan


- Original Message -
> From: "张腾" <2573552...@qq.com>
> To: "bioc-devel" 
> Sent: Thursday, March 23, 2017 7:33:22 PM
> Subject: [Bioc-devel] Unit tests

> Hello all,
> 
> I am not sure whether the Unit tests are necessary to release a package for
> Bioconductor. In my package, only one main function to export and other
> functions are hidden into the main function. In my package, the input data are
> some files in BAM format and the output is a report in HTML format. So I want
> to know if the unit tests are very important and necessary for releasing a
> package. If it is necessary, how can I add these tests to my package in this
> special input and output situation?
> 
> Can anyone help me solve this problem? Thank you very much
>   [[alternative HTML version deleted]]
> 
> ___
> Bioc-devel@r-project.org mailing list
> https://urldefense.proofpoint.com/v2/url?u=https-3A__stat.ethz.ch_mailman_listinfo_bioc-2Ddevel=DwICAg=eRAMFD45gAfqt84VtBcfhQ=TF6f93hjWmgMzjqP9F3thRifibmFvfjc5Ae-bzNwDGo=-y21Sirq2bdhul9Ar8vnDiy_LjASe_ND5ozxWVwudWw=Wroy1VJao_ttDgKZZY7vOApZkTTauP_S8kAgYTs6jo4=

___
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel

Re: [Bioc-devel] Unit Tests & Test Coverage

[Hervé Pagès]

Hi Juan Pablo,

Thanks for adding unit tests to your package. We really encourage
developers to do this. I'm sorry that this is not reflected on
EventPointer's landing page.

I investigated this a bit (I'm not familiar with how the coverage
badges are generated on the package landing pages) and here is
what I found so far.

One possible reason for this is that some packages on
Bioconductor-mirror are lagging behind their latest svn revision.
This is actually the case right now for EventPointer. Latest svn
commit is 127626:

  https://hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/EventPointer

but the package is still at commit 127616 on Bioconductor-mirror:

  https://github.com/Bioconductor-mirror/EventPointer

This is a known issue. We'll try to remedy this ASAP.

This can be a reason why coverage is not updated on a package
landing page. That's because the script we use for computing
the coverage interacts with our svn repo (on hedgehog.fhcrc.org)
in order to get the latest revision of the package, but also with
the Bioconductor-mirror account on github
(https://github.com/Bioconductor-mirror/) in order to map the
latest svn revision of a package to the corresponding git commit
on Bioconductor-mirror. This seems to be needed in order to upload
the coverage results to codecov.io where the coverage history is
kept e.g.


https://codecov.io/github/Bioconductor-mirror/VariantAnnotation/branch/master

The coverage information displayed on a package landing page is coming
from there.

When a package is lagging behind on Bioconductor-mirror, the mapping
between last svn revision and git commit fails and so uploading the
coverage results to codecov.io also fails.

However, in the case of EventPointer, there seems to be already 1 point
in the coverage history at codecov.io:

  https://codecov.io/github/Bioconductor-mirror/EventPointer/branch/master

It shows a coverage of about 30%. For some reason, this doesn't show
up on EventPointer's landing page. I'll investigate more and will
let you know.

H.


On 03/23/2017 04:58 AM, Romero, Juan Pablo wrote:

Hi,


I'm currently doing the last updates to my package EventPointer before the 
deadline for next BioC release.


I've added unit tests according to the guide lines in 
https://urldefense.proofpoint.com/v2/url?u=http-3A__bioconductor.org_developers_unitTesting-2Dguidelines_=DwIFaQ=eRAMFD45gAfqt84VtBcfhQ=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA=4xEe7s13qw8OtBMzGMqJ0vk6PIoU68FR0RJ88otY0yg=47DUypCrCnkcv9C1Mlybb0QHi3RLMZh57Q35XoIXkBU=
  and the package

build report shows no problems for both build and check in all systems. 
However, the test coverage tab at top of the package page

still displays "unknown" and im not sure if there is any problem with the tests 
or how to find possible errors.


Here is what the build report shows in all OS:

* checking examples ... OK
* checking for unstated dependencies in �tests� ... OK
* checking tests ...
  Running �runTests.R�
 OK
* checking for unstated dependencies in vignettes ... OK
* checking package vignettes in �inst/doc� ... OK

Thanks!


Best regards,


Juan Pablo Romero



[[alternative HTML version deleted]]



___
Bioc-devel@r-project.org mailing list
https://urldefense.proofpoint.com/v2/url?u=https-3A__stat.ethz.ch_mailman_listinfo_bioc-2Ddevel=DwICAg=eRAMFD45gAfqt84VtBcfhQ=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA=4xEe7s13qw8OtBMzGMqJ0vk6PIoU68FR0RJ88otY0yg=cT2EjT9BIxY3Q8Nmikk0rGfWteNuwuaDEpwNFwpjL2g=



--
Hervé Pagès

Program in Computational Biology
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N, M1-B514
P.O. Box 19024
Seattle, WA 98109-1024

E-mail: hpa...@fredhutch.org
Phone:  (206) 667-5791
Fax:(206) 667-1319

___
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel

Re: [Bioc-devel] Unit tests present/ coverage "unknown"

[Laurent Gatto]

I do see the same problem for rols. I assume this will clear up at some
point, although any suggestion on how to avoid such glitches would be
helpful.

Laurent

On 25 March 2016 16:42, Zach Skidmore wrote:

> Hi All,
>
> Maybe i'm missing something but I don't quite understand why the "test 
> coverage" badge on the GenVisR page displays unknown. I have unit tests 
> set up with testthat, and looking at the  R CMD CHECK reports I see that 
> they are running however the badge still reads as "unknown". Any Ideas? 
> I have looked through the bioc devel pages to no avail, I assume I have 
> to set a flag somewhere.
>
> Thanks!
> Zach

___
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel

Re: [Bioc-devel] unit tests for C code inside a package

[Elena Grassi]

Thank you both for your suggestions.

Peter: Rcpp attributes seem great but right now I would prefer to
avoid changing all the pure C structs of my old codebase in C++
classes and so on. But if in the future we will need to add new
features I will definitely add these two ++ :)
Dan: yep, that was roughly my idea, I wanted to check if that's reasonable/ok.

E.
-- 
$ pom

___
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel

Re: [Bioc-devel] unit tests for C code inside a package

[Peter Haverty]

My favorite solution to this would be to use Rcpp attributes to add R-level 
functions for each C function. You can use these just for testing and skip the 
exporting and manual pages. That's not quite what you asked for but it would 
work. 
BTW transcription factors are my thing so I'm eager to try out the package. 
Thanks!

Typed with thumbs.

 On Jan 26, 2015, at 1:14 AM, Elena Grassi grass...@gmail.com wrote:
 
 Hi,
 
 I'm writing a package that calculates total affinity (see PMID
 21335606 and 16873464 if transcription factors are your thing):
 up until now in our lab we've used a pure C tool that needs fasta and
 tabular formatted PFM-PWM files but we are willing to produce
 something more comfortable that stems from some related Bioc packages
 (TFBSTools and JASPAR2014 basically).
 The package has two simple R methods that call a C entry point and
 then all the calculations are performed by the C code.
 I'd like to write extended unit tests but I am not sure how to do it:
 I've some tests for the R portions (that obviously depends
 also on the C calculation in some parts) but I would like to test in a
 more fine grained way the C code therefore RUnit and BiocGenerics
 seems to solve only a portion of my problem. I would like to use a
 C-based library for unit testing and link it to the automatic
 check done for the package thanks to BiocGenerics: is this reasonable?
 I've looked at some other packages without being able to find
 something similar to this.
 I've read 
 http://stackoverflow.com/questions/26322135/unit-testing-rcpp-code-in-a-package,
 but I think
 that it would be nice to test the C code with the whole package.
 
 Thanks,
 E.
 ps. the source code right now is here:
 https://github.com/vodkatad/MatrixRider (the vignette is on its way,
 the nodevel branch is the one active now and it works with R version
 3.1.2 and Bioconductor 3.0 for our internal use).
 
 -- 
 $ pom
 
 ___
 Bioc-devel@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/bioc-devel

___
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel

Re: [Bioc-devel] unit tests for C code inside a package

[Dan Tenenbaum]

- Original Message -
 From: Elena Grassi grass...@gmail.com
 To: bioc-devel@r-project.org
 Sent: Monday, January 26, 2015 1:14:59 AM
 Subject: [Bioc-devel] unit tests for C code inside a package
 
 Hi,
 
 I'm writing a package that calculates total affinity (see PMID
 21335606 and 16873464 if transcription factors are your thing):
 up until now in our lab we've used a pure C tool that needs fasta and
 tabular formatted PFM-PWM files but we are willing to produce
 something more comfortable that stems from some related Bioc packages
 (TFBSTools and JASPAR2014 basically).
 The package has two simple R methods that call a C entry point and
 then all the calculations are performed by the C code.
 I'd like to write extended unit tests but I am not sure how to do it:
 I've some tests for the R portions (that obviously depends
 also on the C calculation in some parts) but I would like to test in
 a
 more fine grained way the C code therefore RUnit and BiocGenerics
 seems to solve only a portion of my problem. I would like to use a
 C-based library for unit testing and link it to the automatic
 check done for the package thanks to BiocGenerics: is this
 reasonable?
 I've looked at some other packages without being able to find
 something similar to this.
 I've read
 http://stackoverflow.com/questions/26322135/unit-testing-rcpp-code-in-a-package,
 but I think
 that it would be nice to test the C code with the whole package.
 

Choose a C-based unit-testing library, include it with your package, write a C 
function that calls your test suite, register that function to be callable from 
R, and call it from an R unit test?

Dan


 Thanks,
 E.
 ps. the source code right now is here:
 https://github.com/vodkatad/MatrixRider (the vignette is on its way,
 the nodevel branch is the one active now and it works with R version
 3.1.2 and Bioconductor 3.0 for our internal use).
 
 --
 $ pom
 
 ___
 Bioc-devel@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/bioc-devel


___
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel