[PATCH] r-base suppresses ALL error messages forever

Doh, I found the issue:

in /etc/Rprofile.site you first save options in variable ow, turn off error messages and then check for Revolution-R stuff. Later you restore the old options by reapplying the "ow" options again. However "ow" only contains the "warn" option but you also modify "show.error.messages"

ow <- options("warn")
options("warn"=-1,show.error.messages=FALSE)
[...]
options(ow)

I solved the issue by also saving option "show.error.message", that is by applying:

-ow <- options("warn")
+ow <- options("warn","show.error.message")

to /etc/Rprofile.site

I have no /etc/Rprofile.site file, but i think you meant /etc/R/Rprofile.site, note the extra "R" directory.

secondly, the change should have added the "s" to "messages".

-ow <- options("warn")
+ow <- options("warn","show.error.messages")

after these changes, i can confirm this bug and the fix.

However a simpler solution is to just store ALL of the options by making the following changes.

-ow <- options("warn")
+ow <- options()

both of these changes filled this incredibly annoying problem for me.

on further investigation i think the

-ow <- options("warn")
+ow <- options()

patch is a bad idea because the "require(Revobase, quietly=TRUE)" adds 10 addition options to "options()" and Revobase may currently or in the future change pre-existing options, which would be overwritten with "options(ow)". so i think the better solution is...

-ow <- options("warn")
+ow <- options("warn","show.error.messages")

Thanks for the bug report and fix -- I concur. This was an oversight, and we will apply the promptly.

Thanks also to whoever subscribed me to this today....

Dirk

Thanks for the bug report and fix -- I concur. This was an oversight, and we will apply the fix promptly.

Thanks also to whoever subscribed me to this today....

Dirk

@jbcolvin: thanks for correcting my typos, I was typing from memory. Yes, your changes were exactly what I had intended.

@Dirk, glad I could help out, although it took me 2 days until I found the culprit. :-)

Hum.... this but does not have a patch attached.

no patch as file, but it's a one liner. see: https://bugs.launchpad.net/ubuntu/+source/r-base/+bug/426360/comments/1

inlined "patch" (/etc/R/Rprofile.site):

-ow <- options("warn")
+ow <- options("warn","show.error.messages")

On 18 September 2009 at 09:19, spaetz wrote:
| no patch as file, but it's a one liner. see:
| https://bugs.launchpad.net/ubuntu/+source/r-base/+bug/426360/comments/1
|
|
| inlined "patch" (/etc/R/Rprofile.site):
|
| -ow <- options("warn")
| +ow <- options("warn","show.error.messages")

Correct. I directly modified the file debian/Rprofile.site which debian/rules
copies out.

Thanks again for the bug report and help! Took a little longer to get it out
as it got wrapped into two other issues related to the REvo R packages. It
should all work fine now -- as it does on my amd64 box using these packages.

Dirk

| --
| [PATCH] r-base suppresses ALL error messages forever
| https://bugs.launchpad.net/bugs/426360
| You received this bug notification because you are a direct subscriber
| of the bug.

--
Three out of two people have difficulties with fractions.

Dirk, do you have a package somewhere that fixes this? Your comments above inclined me to think that you did, but I don't know where to look. I'd be happy to guide that into Karmic, or to simply apply the fix given by spaetz if you say the word.

Hi Colin,

Thanks for following-up!

On 25 September 2009 at 21:55, Colin Watson wrote:
| Dirk, do you have a package somewhere that fixes this? Your comments
| above inclined me to think that you did, but I don't know where to look.

Sorry -- didn't publicize that all that well. It is in the 'revor' PPA:

      https://launchpad.net/~revor/+archive/ppa

| I'd be happy to guide that into Karmic, or to simply apply the fix given
| by spaetz if you say the word.

It should already be in Karmic. I tend to do the package work, Scott R does
the uploads and Scott K approves them.

This was egg-on-face bug originating with our REvolution R extensions that I
should have caught but didn't. But it should be fixed already, plus we have
another upload slated.

Cheers, Dirk

|
| --
| [PATCH] r-base suppresses ALL error messages forever
| https://bugs.launchpad.net/bugs/426360
| You received this bug notification because you are a direct subscriber
| of the bug.

--
Three out of two people have difficulties with fractions.

On Fri, Sep 25, 2009 at 05:14:19PM -0500, Dirk Eddelbuettel wrote:
> On 25 September 2009 at 21:55, Colin Watson wrote:
> | Dirk, do you have a package somewhere that fixes this? Your comments
> | above inclined me to think that you did, but I don't know where to look.
>
> Sorry -- didn't publicize that all that well. It is in the 'revor' PPA:
>
> https://launchpad.net/~revor/+archive/ppa
>
> | I'd be happy to guide that into Karmic, or to simply apply the fix given
> | by spaetz if you say the word.
>
> It should already be in Karmic. I tend to do the package work, Scott R does
> the uploads and Scott K approves them.

It's not yet in Karmic; that's why I asked. :-) (It's not in any
approval queue either, as far as I can see.)

You left debian/changelog~ and debian/Rprofile.site~ lying around in the
diff - presumably unintentional? I removed it, and uploaded your package
otherwise unmodified except for changelog updates. I'd like to get this
fixed for the 9.10 beta, hence the rush - Scott R, I hope I'm not
stepping on your toes.

Although I didn't change it for this upload since I wanted to discuss it
with you, I think that for Karmic it would be a good idea to put the
dependencies back the way they were before 2.9.2-2ubuntu1~ppa2, even if
they need to stay otherwise in your PPA; do you disagree? In particular,
if the dpkg dependencies are left the way they are at the moment then it
will be possible for partial upgrades to go wrong, AFAICS. I'd rather
give correct dependencies in Karmic precedence over installability on
Jaunty, at least for packages in the main archive.

Thanks,

--
Colin Watson [<email address hidden>]

This bug was fixed in the package r-base - 2.9.2-2ubuntu1

---------------
r-base (2.9.2-2ubuntu1) karmic; urgency=low

  * Update from https://launchpad.net/~revor/+archive/ppa.

r-base (2.9.2-2ubuntu1~ppa3) karmic; urgency=low

  * etc/R/Rprofile.site: Improved test for when to be 'quiet'

r-base (2.9.2-2ubuntu1~ppa2) karmic; urgency=low

  * debian/control: Soften changes to keep package installable on Jaunty
    - revert Builds-Depends to libreadline5-dev
    - revert r-base-dev Depends to libreadline5-dev
    - revert r-doc-info Depends to unversioned dpkg

r-base (2.9.2-2ubuntu1~ppa1) karmic; urgency=low

  * Synced from Debian unstable

  * debian/control: As requested by Matthias Klose, change
    Build-Depends: from 'libreadline5-dev' to 'libreadline-dev'

  * debian/control: Set Standards-Version: to current version

  * debian/control: Add Depends: for r-doc-info on
    'dpkg (>= 1.15.4) | install-info'

  * debian/rules: No longer install /etc/ld.so.conf.d/libR.conf

  * debian/r-base-core.postinst: Call paperconf without prepended path

r-base (2.9.2-1ubuntu1~ppa5) karmic; urgency=low

  * etc/R/Rprofile.site: Fixed warning() issue (LP: #426360)
  * etc/R/Rprofile.site: Modified REvolution Computing greetings

r-base (2.9.2-1ubuntu1~ppa4) karmic; urgency=low

  * etc/R/Rprofile.site: Fixed warning() issue (LP: #426360)
  * etc/R/Rprofile.site: Modified REvolution Computing greetings

  * debian/rules: Soft link from /usr/lib/R/COPYING against GPL-2
    instead of GPL, with thanks to Isaac for spotting this.

 -- Colin Watson <email address hidden> Fri, 25 Sep 2009 23:27:58 +0100

On 25 September 2009 at 22:35, Colin Watson wrote:
| On Fri, Sep 25, 2009 at 05:14:19PM -0500, Dirk Eddelbuettel wrote:
| > On 25 September 2009 at 21:55, Colin Watson wrote:
| > | Dirk, do you have a package somewhere that fixes this? Your comments
| > | above inclined me to think that you did, but I don't know where to look.
| >
| > Sorry -- didn't publicize that all that well. It is in the 'revor' PPA:
| >
| > https://launchpad.net/~revor/+archive/ppa
| >
| > | I'd be happy to guide that into Karmic, or to simply apply the fix given
| > | by spaetz if you say the word.
| >
| > It should already be in Karmic. I tend to do the package work, Scott R does
| > the uploads and Scott K approves them.
|
| It's not yet in Karmic; that's why I asked. :-) (It's not in any
| approval queue either, as far as I can see.)

Ahh. Isaac had checked the Alpha builds and noticed that too.

| You left debian/changelog~ and debian/Rprofile.site~ lying around in the
| diff - presumably unintentional? I removed it, and uploaded your package

Yes, my source-only build invocation must have missed the pre-clean switch.

| otherwise unmodified except for changelog updates. I'd like to get this
| fixed for the 9.10 beta, hence the rush - Scott R, I hope I'm not
| stepping on your toes.

Great, we appreciate the effort of getting it into the 9.10 beta! It should
be there.

| Although I didn't change it for this upload since I wanted to discuss it
| with you, I think that for Karmic it would be a good idea to put the
| dependencies back the way they were before 2.9.2-2ubuntu1~ppa2, even if
| they need to stay otherwise in your PPA; do you disagree? In particular,
| if the dpkg dependencies are left the way they are at the moment then it
| will be possible for partial upgrades to go wrong, AFAICS. I'd rather
| give correct dependencies in Karmic precedence over installability on
| Jaunty, at least for packages in the main archive.

Concretely, we are talking about
 - support for libreadline-dev (to get libreadline6)
 - support for dpkg triggers (hence the dpkg >= requirement)

I agree. My short-term convenience should not matter. I can always rebuild on
Jaunty.

So that would mean another ~ppa$X upload, right?

Dirk, out of town at the kid's soccer tournament but with ssh access

| Thanks,
|
| --
| Colin Watson [<email address hidden>]
|
| --
| [PATCH] r-base suppresses ALL error messages forever
| https://bugs.launchpad.net/bugs/426360
| You received this bug notification because you are a direct subscriber
| of the bug.

--
Three out of two people have difficulties with fractions.

On 25 September 2009 at 21:34, Dirk Eddelbuettel wrote:
| On 25 September 2009 at 22:35, Colin Watson wrote:
| | On Fri, Sep 25, 2009 at 05:14:19PM -0500, Dirk Eddelbuettel wrote:
| | > On 25 September 2009 at 21:55, Colin Watson wrote:
| | Although I didn't change it for this upload since I wanted to discuss it
| | with you, I think that for Karmic it would be a good idea to put the
| | dependencies back the way they were before 2.9.2-2ubuntu1~ppa2, even if
| | they need to stay otherwise in your PPA; do you disagree? In particular,
| | if the dpkg dependencies are left the way they are at the moment then it
| | will be possible for partial upgrades to go wrong, AFAICS. I'd rather
| | give correct dependencies in Karmic precedence over installability on
| | Jaunty, at least for packages in the main archive.
|
| Concretely, we are talking about
| - support for libreadline-dev (to get libreadline6)
| - support for dpkg triggers (hence the dpkg >= requirement)
|
| I agree. My short-term convenience should not matter. I can always rebuild on
| Jaunty.
|
| So that would mean another ~ppa$X upload, right?
|
| Dirk, out of town at the kid's soccer tournament but with ssh access

A new build (r-base_2.9.2-2ubuntu1~ppa4_source.changes) just went to the PPA,
sole change was to re-enable the current Debian Policy / Karmic changes for
the libreadline-dev switch and the versioned dpkg Depends to rely on triggers.

Source upload, so no .deb packages here and no debdiff to attach. Sorry.

Scott R and Scott K:
      If you could do you two-step once more 2.9.2-2ubuntu1~ppa4 then we
      we should be in great shape.

Dirk

--
Three out of two people have difficulties with fractions.