Hello, It may have been a long time without it but still...https://ci.inria.fr/pharo-contribution/job/Pillar/PHARO=30,VERSION=stable,VM=vm/lastSuccessfulBuild/artifact/Pillar.zip I'll go to 4.0 but there are other dependencies etc for the OS maybe etc. Not good as I need to migrate these things for my customer as well. |
Le 30/09/2015 18:38, [hidden email] a écrit :
> Hello, > > It may have been a long time without it but still... > > I ran the doc job on my CI and got this that wasn't working: > > https://ci.inria.fr/pharo-contribution/job/Pillar/PHARO=30,VERSION=stable,VM=vm/lastSuccessfulBuild/artifact/Pillar.zip > > Hum, yes, but ... why remove the doc tools as this is quite an important > thing to keep around for all Pharo versions? > > I'll go to 4.0 but there are other dependencies etc for the OS maybe etc. > Not good as I need to migrate these things for my customer as well. > > Phil We removed the Pillar job for Pharo 3 because we introduced an new feature not backward compatible. So we freezed the Pillar version for Pharo 3. So we do not need a job for it anymore since it will not change and we do not want to keep a red job for "Pillar - Pharo 3 - Development". You can still load the stable version of Pillar in Pharo 3, but this version will not evolve anymore. -- Cyril Ferlicot http://www.synectique.eu 165 Avenue Bretagne Lille 59000 France signature.asc (836 bytes) Download Attachment |
Ok, where? Phil On Wed, Sep 30, 2015 at 7:57 PM, Ferlicot D. Cyril <[hidden email]> wrote: Le 30/09/2015 18:38, [hidden email] a écrit : |
In reply to this post by CyrilFerlicot
Downloaded the book-skeleton, patched the download script so that it gets the centos things right and the Pharo4VM.
Now, changes in Pillar made my books unsuable, namely the introduction of things like this: ${tag:parameter=value|parameter2=value2}$ made contents with bash shell scripts unparsable. Yeah, but there are often such things. Like in: Il existe une autre variable ==${ACQSYS_RRDPRINT_TIMEFRAME_SHORT}== qui couvre une période plus courte How do I solve that with the new markup? Phil On Wed, Sep 30, 2015 at 7:57 PM, Ferlicot D. Cyril <[hidden email]> wrote: > > Le 30/09/2015 18:38, [hidden email] a écrit : > > Hello, > > > > It may have been a long time without it but still... > > > > I ran the doc job on my CI and got this that wasn't working: > > > > https://ci.inria.fr/pharo-contribution/job/Pillar/PHARO=30,VERSION=stable,VM=vm/lastSuccessfulBuild/artifact/Pillar.zip > > > > Hum, yes, but ... why remove the doc tools as this is quite an important > > thing to keep around for all Pharo versions? > > > > I'll go to 4.0 but there are other dependencies etc for the OS maybe etc. > > Not good as I need to migrate these things for my customer as well. > > > > Phil > > Hi Phil, > > We removed the Pillar job for Pharo 3 because we introduced an new > feature not backward compatible. > So we freezed the Pillar version for Pharo 3. So we do not need a job > for it anymore since it will not change and we do not want to keep a red > job for "Pillar - Pharo 3 - Development". > > You can still load the stable version of Pillar in Pharo 3, but this > version will not evolve anymore. > > -- > Cyril Ferlicot > > http://www.synectique.eu > > 165 Avenue Bretagne > Lille 59000 France > |
Le 30/09/2015 22:32, [hidden email] a écrit :
> Downloaded the book-skeleton, patched the download script so that it > gets the centos things right and the Pharo4VM. > > Now, changes in Pillar made my books unsuable, namely the introduction > of things like this: > > ${tag:parameter=value|parameter2=value2}$ > > > made contents with bash shell scripts unparsable. > > > Yeah, but there are often such things. > > > Like in: > > > Il existe une autre variable ==${ACQSYS_RRDPRINT_TIMEFRAME_SHORT}== qui > couvre une période plus courte > > > How do I solve that with the new markup? > > > Phil > > > Il existe une autre variable ==\${ACQSYS_RRDPRINT_TIMEFRAME_SHORT}== qui couvre une période plus courte :) -- Cyril Ferlicot http://www.synectique.eu 165 Avenue Bretagne Lille 59000 France signature.asc (836 bytes) Download Attachment |
Which is a pain as when I copy/paste a shell script full of those, I won't do that. Is this also doing the same thing in blocks like:[[[label=.bashrc|caption=Ajouts à .bashrc|language=bash ]]] I am not gonna escape those things as what happens is copy/pasting code. That there are no such directives in Markdiwn is what makes it nice to use and hassle free. I would escape your directives and leave the core of the text alone. Well, what do you think? Phil On Wed, Sep 30, 2015 at 10:38 PM, Ferlicot D. Cyril <[hidden email]> wrote: Le 30/09/2015 22:32, [hidden email] a écrit : |
In reply to this post by CyrilFerlicot
On 30-09-15 19:57, Ferlicot D. Cyril wrote:
> We removed the Pillar job for Pharo 3 because we introduced an new > feature not backward compatible. Why is that a reason to remove the build? You have a configuration, don't you? Stephan |
On 01-10-15 09:41, Stephan Eggermont wrote:
> On 30-09-15 19:57, Ferlicot D. Cyril wrote: > >> We removed the Pillar job for Pharo 3 because we introduced an new >> feature not backward compatible. > > Why is that a reason to remove the build? You have a configuration, > don't you? I would suggest: move to a separate job that is run much less often and doesn't keep so many old artifacts. That makes it easier to get old stuff running again. Stephan |
In Jenkins you specify configuration matrix (versions of pharo versus type (development, stable, ...)).. So it will run development for all versions, even pharo 30 (which is now permanently broken) From: [hidden email] Sent: 10/1/2015 9:47 AM To: [hidden email] Subject: Re: [Pharo-users] Pillar for 30 gone the way of the dodo on the CI > On 30-09-15 19:57, Ferlicot D. Cyril wrote: > >> We removed the Pillar job for Pharo 3 because we introduced an new >> feature not backward compatible. > > Why is that a reason to remove the build? You have a configuration, > don't you? I would suggest: move to a separate job that is run much less often and doesn't keep so many old artifacts. That makes it easier to get old stuff running again. Stephan |
Is CI an archive?
|
What I mean: it seems to me not possible to keep all CI build alive “forever” as a CI server build. Shouldn’t we separate “archival of builds” from “active ci”? For Pharo itself we do that, as we do not have the manpower (we just don’t) to keep the CI alive for stable and dev, yet every build ever is on files.pharo.org *forever*, archived. Marcus
|
In reply to this post by philippeback
[hidden email] <[hidden email]> writes: > Ok, where? in Pharo 3, open the Catalog Browser, choose Pillar, and click "Install Stable Version". You will get 2 warnings you can safely ignore. You can run the unit tests to check everything works fine (Pillar contains 2198 unit tests, in this version, covering 92% of the source code). -- Damien Cassou http://damiencassou.seasidehosting.st "Success is the ability to go from one failure to another without losing enthusiasm." --Winston Churchill |
In reply to this post by philippeback
[hidden email] <[hidden email]> writes: > Downloaded the book-skeleton, patched the download script so that it gets > the centos things right and the Pharo4VM. > > Now, changes in Pillar made my books unsuable, namely the introduction of > things like this: > > ${tag:parameter=value|parameter2=value2}$ > > > made contents with bash shell scripts unparsable. indeed, sorry about that. This new syntax breaks existing documents. Nevertheless, this new syntax has been created exactly for the purpose of limiting future problems: new syntax will not be necessary most of the time because ${tag}$ is pluggable. -- Damien Cassou http://damiencassou.seasidehosting.st "Success is the ability to go from one failure to another without losing enthusiasm." --Winston Churchill |
In reply to this post by philippeback
[hidden email] <[hidden email]> writes: > Which is a pain as when I copy/paste a shell script full of those, I won't > do that. I agree that's a pain. But you have to do it. > Is this also doing the same thing in blocks like: > > [[[label=.bashrc|caption=Ajouts à .bashrc|language=bash > > ${SOMEVAR:-} > > ]]] text between [[[ and ]]] is used verbatim and thus not parsed by Pillar. You can safely use Pillar-protected characters in there and they won't be interpreted. > I am not gonna escape those things as what happens is copy/pasting code. > That there are no such directives in Markdiwn is what makes it nice to use > and hassle free. this is also what makes Markdown less powerful. We had to find a compromise. I think the ${tag}$ is a good one because it is pluggable. We can use it for many things: file inclusion, table of content declaration, bibliography citations, any plugin you can think of. > I would escape your directives and leave the core of the text alone. I didn't get that. -- Damien Cassou http://damiencassou.seasidehosting.st "Success is the ability to go from one failure to another without losing enthusiasm." --Winston Churchill |
In reply to this post by Peter Uhnak
On 01-10-15 10:05, Peter Uhnak wrote:
> In Jenkins you specify configuration matrix (versions of pharo versus type (development, stable, ...)).. So it will run development for all versions, even pharo 30 (which is now permanently broken) Shouldn't #'development' for Pharo3.x then point to #'stable'? There is no development, there could be bugfixes. The matrix indeed assumes that you want all combinations, while what we actually need for the build is something like the Matrix Combinations Plugin https://wiki.jenkins-ci.org/display/JENKINS/Matrix+Combinations+Plugin I haven't tested the plugin yet. Stephan |
You can exclude some of the combinations from matrix. E.g. exclude build for 30 & development.
> On 01 Oct 2015, at 14:04, Stephan Eggermont <[hidden email]> wrote: > > On 01-10-15 10:05, Peter Uhnak wrote: >> In Jenkins you specify configuration matrix (versions of pharo versus type (development, stable, ...)).. So it will run development for all versions, even pharo 30 (which is now permanently broken) > > Shouldn't #'development' for Pharo3.x then point to #'stable'? > There is no development, there could be bugfixes. > > The matrix indeed assumes that you want all combinations, > while what we actually need for the build is something like > the Matrix Combinations Plugin > https://wiki.jenkins-ci.org/display/JENKINS/Matrix+Combinations+Plugin > I haven't tested the plugin yet. > > Stephan > > > |
In reply to this post by Stephan Eggermont-3
Stephan Eggermont <[hidden email]> writes: > On 01-10-15 10:05, Peter Uhnak wrote: >> In Jenkins you specify configuration matrix (versions of pharo versus type (development, stable, ...)).. So it will run development for all versions, even pharo 30 (which is now permanently broken) > > Shouldn't #'development' for Pharo3.x then point to #'stable'? > There is no development, there could be bugfixes. > > The matrix indeed assumes that you want all combinations, > while what we actually need for the build is something like > the Matrix Combinations Plugin > https://wiki.jenkins-ci.org/display/JENKINS/Matrix+Combinations+Plugin > I haven't tested the plugin yet. I don't want the main Pillar job to be marked as having failed for a platform I don't want to maintain. Please create a new job as you proposed in another email. Nevertheless, I agree with Marcus that continuous integration should not be used as an image archive. -- Damien Cassou http://damiencassou.seasidehosting.st "Success is the ability to go from one failure to another without losing enthusiasm." --Winston Churchill |
In reply to this post by Stephan Eggermont-3
>>> We removed the Pillar job for Pharo 3 because we introduced an new >>> feature not backward compatible. >> >> Why is that a reason to remove the build? You have a configuration, >> don't you? > > I would suggest: move to a separate job that is run much less often > and doesn't keep so many old artifacts. That makes it easier to get > old stuff running again. + 10000 |
In reply to this post by Marcus Denker-4
So let us archive it but now my scripts are broken.
Stef Le 1/10/15 10:10, Marcus Denker a
écrit :
|
On 03-10-15 09:16, stepharo wrote:
> So let us archive it but now my scripts are broken. If we want the old scripts to continue working, instead of creating a new (old) job for it we would need some mechanism to tell the build matrix in more detail what should happen to which build: only build nightly/weekly, and keep just a few artifacts Stephan |
Free forum by Nabble | Edit this page |