GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Kolen Cheung]

[-- Attachment #1.1: Type: text/plain, Size: 1910 bytes --]



Hi,

I’m currently experimenting with Travis CI build against *all* pandoc 
versions. The script I use:

env:
  - pandocVersion=1.13.2
  - pandocVersion=1.14
  - pandocVersion=1.14.0.1
  - pandocVersion=1.14.1
  - pandocVersion=1.15
  - pandocVersion=1.15.0.5
  - pandocVersion=1.15.0.6
  - pandocVersion=1.15.1
  - pandocVersion=1.15.2
  - pandocVersion=1.16.0.1
  - pandocVersion=1.16.0.2
  - pandocVersion=1.17
  - pandocVersion=1.17.0.1
  - pandocVersion=1.17.0.2
  - pandocVersion=1.17.2
  - pandocVersion=1.18
before_install:
  - wget https://github.com/jgm/pandoc/releases/download/$pandocVersion/pandoc-$pandocVersion-1-amd64.deb &&
    sudo dpkg -i pandoc-$pandocVersion-1-amd64.deb

The key to make this possible is the pattern of the filename of the .deb 
always end in -1-amd64.deb. There are a few instances that this is not 
true, the newest version among them is pandoc-1.17.1-2-amd64.deb. (The 
script above includes all pandoc versions that follows the pattern.)

To be clear, I am not suggesting we go back to fix the releases that has 
been released. The purpose of this thread is only to

   1. see if this has been fixed already 
   2. if not fixed, how to prevent this to happen in the future 
   3. perhaps to be curious about why this would occur 

Thanks.
​

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/788928f7-3cd9-4edc-bcc4-c80e8a75a5df%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.2: Type: text/html, Size: 9038 bytes --]

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[John MacFarlane]

It's not a problem that needs to be fixed.  The -2 suffix is
used instead of -1 when there was a prior, flawed -1 release
that needed to be fixed.

+++ Kolen Cheung [Nov 13 16 15:30 ]:
>   Hi,
>
>   I’m currently experimenting with Travis CI build against all pandoc
>   versions. The script I use:
>env:
>  - pandocVersion=1.13.2
>  - pandocVersion=1.14
>  - pandocVersion=1.14.0.1
>  - pandocVersion=1.14.1
>  - pandocVersion=1.15
>  - pandocVersion=1.15.0.5
>  - pandocVersion=1.15.0.6
>  - pandocVersion=1.15.1
>  - pandocVersion=1.15.2
>  - pandocVersion=1.16.0.1
>  - pandocVersion=1.16.0.2
>  - pandocVersion=1.17
>  - pandocVersion=1.17.0.1
>  - pandocVersion=1.17.0.2
>  - pandocVersion=1.17.2
>  - pandocVersion=1.18
>before_install:
>  - wget https://github.com/jgm/pandoc/releases/download/$pandocVersion/pandoc-$
>pandocVersion-1-amd64.deb &&
>    sudo dpkg -i pandoc-$pandocVersion-1-amd64.deb
>
>   The key to make this possible is the pattern of the filename of the
>   .deb always end in -1-amd64.deb. There are a few instances that this is
>   not true, the newest version among them is pandoc-1.17.1-2-amd64.deb.
>   (The script above includes all pandoc versions that follows the
>   pattern.)
>
>   To be clear, I am not suggesting we go back to fix the releases that
>   has been released. The purpose of this thread is only to
>    1. see if this has been fixed already
>    2. if not fixed, how to prevent this to happen in the future
>    3. perhaps to be curious about why this would occur
>
>   Thanks.
>   ​
>
>   --
>   You received this message because you are subscribed to the Google
>   Groups "pandoc-discuss" group.
>   To unsubscribe from this group and stop receiving emails from it, send
>   an email to [1]pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
>   To post to this group, send email to
>   [2]pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
>   To view this discussion on the web visit
>   [3]https://groups.google.com/d/msgid/pandoc-discuss/788928f7-3cd9-4edc-
>   bcc4-c80e8a75a5df%40googlegroups.com.
>   For more options, visit [4]https://groups.google.com/d/optout.
>
>References
>
>   1. mailto:pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org
>   2. mailto:pandoc-discuss-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org
>   3. https://groups.google.com/d/msgid/pandoc-discuss/788928f7-3cd9-4edc-bcc4-c80e8a75a5df-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org?utm_medium=email&utm_source=footer
>   4. https://groups.google.com/d/optout

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/20161115133045.GB4589%40Administrateurs-iMac-3.local.
For more options, visit https://groups.google.com/d/optout.

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Kolen Cheung]

[-- Attachment #1: Type: text/plain, Size: 337 bytes --]

I see. But since it is a fix, it seems to me the last digit of the semantic versioning should be incremented instead.

And I also want to understand why the original -1 release needed to be fix. I am guessing they are generated by Travis and the -2 is some manual adjustment that make it work (while the one by Travis doesn't)?

Thnaks.

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Ivan Lazar Miljenovic]

My guess is that there's a difference between a .deb package issue
that needs to be fixed and re-released, and an actual code issue (that
affects _all_ packages, Hackage releases, etc.) which would require a
semantic versioning increment.

On 16 November 2016 at 10:36, Kolen Cheung <christian.kolen-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> I see. But since it is a fix, it seems to me the last digit of the semantic versioning should be incremented instead.
>
> And I also want to understand why the original -1 release needed to be fix. I am guessing they are generated by Travis and the -2 is some manual adjustment that make it work (while the one by Travis doesn't)?
>
> Thnaks.
>
> --
> You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
> To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
> To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/fb9b925e-638f-4ef3-8ecf-682e4cc3079a%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.



-- 
Ivan Lazar Miljenovic
Ivan.Miljenovic-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org
http://IvanMiljenovic.wordpress.com

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Kolen Cheung]

[-- Attachment #1: Type: text/plain, Size: 1394 bytes --]

I understand that, and that's probably the reason of the existance of the last `-1` versioning. But that breaks the pattern to script it automatically. Again, the historical packages are not important, but let say there's a latest 18.1 released, then the latest .deb actually becomes 18.1-2... Then either I break the script pattern for the latest version (and what if I need to test against 17.2 and 18.1, where both of them ends with a different -1 -2?), or I hold back to not use the latest version (so I didn't test against a newer verion colleagues may be using).

So from the scripting perspective, it seems even if only one of the platform needs a bug fix, the last digit of *all* of them should be bumped up while keeping -1 appended. -1 is still needed because older version of pandoc has that pattern.

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/e4d6fa41-dcca-41b3-b47e-b8057f5dfec4%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Ivan Lazar Miljenovic]

But why should anyone using a packaging system other than .deb have
random version bumps just because someone didn't build a .deb
correctly (or needs to do an update due to system libraries, etc.)?

This may even be a Debian packaging requirement (I wouldn't know; the
closest I've had to dealing with Debian-derived distros is Ubuntu for
Travis-CI and Docker).

On 16 November 2016 at 11:39, Kolen Cheung <christian.kolen-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> I understand that, and that's probably the reason of the existance of the last `-1` versioning. But that breaks the pattern to script it automatically. Again, the historical packages are not important, but let say there's a latest 18.1 released, then the latest .deb actually becomes 18.1-2... Then either I break the script pattern for the latest version (and what if I need to test against 17.2 and 18.1, where both of them ends with a different -1 -2?), or I hold back to not use the latest version (so I didn't test against a newer verion colleagues may be using).
>
> So from the scripting perspective, it seems even if only one of the platform needs a bug fix, the last digit of *all* of them should be bumped up while keeping -1 appended. -1 is still needed because older version of pandoc has that pattern.
>
> --
> You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
> To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
> To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/e4d6fa41-dcca-41b3-b47e-b8057f5dfec4%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.



-- 
Ivan Lazar Miljenovic
Ivan.Miljenovic-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org
http://IvanMiljenovic.wordpress.com

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/CA%2Bu6gbx82%2BK45v2h6DR1eXoVyd1aY6QuY02MhkBeZYiuzigStw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Kolen Cheung]

[-- Attachment #1.1: Type: text/plain, Size: 1212 bytes --]



I have a related question: I am experimenting with speeding up the pandoc 
travis build, so I try to understand more on how it is built.

I find that there’s make_travis_yml.hs and make .travis.yml command in the 
makefile. But the result generated from make .travis.yml is very different 
from the .travis.yml currently used. Is the current travis build hand-tuned 
or using make_travis_yml.hs? In other words, if I want to make change to 
.travis.yml (and potentially make pull request if there’s any speed boost), 
where should I start? make_travis_yml.hs and makefile, or .travis.yml 
directly?

Thanks.
​

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/41de9a9f-b0ee-4b74-8673-1df335d611f7%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.2: Type: text/html, Size: 8934 bytes --]

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Kolen Cheung]

[-- Attachment #1.1: Type: text/plain, Size: 1552 bytes --]



For example, looking at the History for .travis.yml - jgm/pandoc 
<https://github.com/jgm/pandoc/commits/master/.travis.yml>, it seems @jgm 
hand-tuned it.

(Sorry to split it into 2 emails. I should have mentioned it together.)

On Tuesday, November 15, 2016 at 8:00:02 PM UTC-8, Kolen Cheung wrote:

I have a related question: I am experimenting with speeding up the pandoc 
> travis build, so I try to understand more on how it is built.
>
> I find that there’s make_travis_yml.hs and make .travis.yml command in 
> the makefile. But the result generated from make .travis.yml is very 
> different from the .travis.yml currently used. Is the current travis 
> build hand-tuned or using make_travis_yml.hs? In other words, if I want 
> to make change to .travis.yml (and potentially make pull request if 
> there’s any speed boost), where should I start? make_travis_yml.hs and 
> makefile, or .travis.yml directly?
>
> Thanks.
> ​
>
​

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/7835ca9f-5488-483c-824a-cbf78fe62900%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.2: Type: text/html, Size: 19722 bytes --]

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[John MacFarlane]

+++ Kolen Cheung [Nov 15 16 15:36 ]:
>I see. But since it is a fix, it seems to me the last digit of the semantic versioning should be incremented instead.

No, because the semantic versioning refers to the software.

In this case the software was exactly the same; all that
changed was the packaging.  (There had been a problem in
creating the deb originally, so I recreated the deb and
released with the -2 suffix.)

This really shouldn't be a big deal.  Your script just
needs a bit more intelligence.

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[John MacFarlane]

+++ Ivan Lazar Miljenovic [Nov 16 16 12:19 ]:
>But why should anyone using a packaging system other than .deb have
>random version bumps just because someone didn't build a .deb
>correctly (or needs to do an update due to system libraries, etc.)?
>
>This may even be a Debian packaging requirement (I wouldn't know; the
>closest I've had to dealing with Debian-derived distros is Ubuntu for
>Travis-CI and Docker).

Yes, debian always has this versioning on the package, which
is independent of the upstream version.

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

['Jakob Voss' via pandoc-discuss]

[-- Attachment #1.1: Type: text/plain, Size: 816 bytes --]

Hi Kolen,

> I’m currently experimenting with Travis CI build against *all* pandoc 
versions. The script I use:

Interesting, I was also thinking how to best do this. Where can we find 
your full script and repository?

Jakob

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/4bc27ca1-a2b5-490b-ac94-2e25a70c6b09%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.2: Type: text/html, Size: 1236 bytes --]

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Kolen Cheung]

[-- Attachment #1.1: Type: text/plain, Size: 1820 bytes --]

I see. I took your advice to change the scripting logic. I now understand 
this is a non-issue.

But it seems it is still difficult to script for dual-OSes. I'm not 
planning to do it now, but in principle I can setup a matrix that has 2 OS 
version (macOS/Linux) then lists all available pandoc versions. In this 
case if the macOS and Linux version differs (say, pandoc-1.17.1-2-amd64.deb 
and pandoc-1.17.1-1-osx.pkg), it seems the only way to do it is either 
taking away the benefits of "matrix expansion" (say, 2X10 builds rather 
than 20), or write a loop that iterate the last digit until successful 
download.

On Wednesday, November 16, 2016 at 3:26:44 AM UTC-8, John MacFarlane wrote:
>
> +++ Ivan Lazar Miljenovic [Nov 16 16 12:19 ]: 
> >But why should anyone using a packaging system other than .deb have 
> >random version bumps just because someone didn't build a .deb 
> >correctly (or needs to do an update due to system libraries, etc.)? 
> > 
> >This may even be a Debian packaging requirement (I wouldn't know; the 
> >closest I've had to dealing with Debian-derived distros is Ubuntu for 
> >Travis-CI and Docker). 
>
> Yes, debian always has this versioning on the package, which 
> is independent of the upstream version. 
>

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/fe2ffb26-71af-4b01-97f1-406a47e5ba6d%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.2: Type: text/html, Size: 2452 bytes --]

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Kolen Cheung]

[-- Attachment #1.1: Type: text/plain, Size: 3191 bytes --]



Hi, Jakob,

It initially started as a pull-request to panflute to test against 
different pandoc versions. We finally decided to test against 1.17.2 and 
1.18 only.

My latest version that runs through *all* available .deb from GitHub 
release is in panflute/.travis.yml at pandoc-multi-version · ickc/panflute 
<https://github.com/ickc/panflute/blob/pandoc-multi-version/.travis.yml>. 
You can see the Travis job in Build #17 - ickc/panflute - Travis CI 
<https://travis-ci.org/ickc/panflute/builds/176546777>.

It is very interesting to see this 7X18 = 126 builds (Python X pandoc 
versions) running concurrently (actually Travis cap it to about 30~40 jobs 
currently). But it will took about 1/2 to 1 hour to complete it (the number 
from Travis is inaccurate, which is a known issue).

Personally I might do something like this in the “stable” branch that’s not 
updated often. I think it is nice to have Travis checking that your filter 
(or whatever other pandoc workflow) works for all those versions, or 
knowing that a particular pandoc version is a trouble maker for debugging 
purpose.

The final version submitted to panflute is in panflute/.travis.yml at 
master · sergiocorreia/panflute 
<https://github.com/sergiocorreia/panflute/blob/master/.travis.yml>, it 
took below 3 minutes to finish the test: Build #31 - sergiocorreia/panflute 
- Travis CI <https://travis-ci.org/sergiocorreia/panflute/builds/176527618>. 
(one of the key to speed up is by allow_failure. It “finishes” the build 
(green-light you to merge) without waiting for allow_failure.)

I might be writing an article in pandoc wiki about setting up Travis for 
pandoc project. For example, after a lots of trial and error, I can compile 
pandoc documents using Travis’s Linux/macOS build in about 2 minutes. 
(including TeXLive, pandoc, pandocfilters, etc.) It is not as trivial as I 
originally thought. Some naive settings will leads to ~10 minutes build 
time. (timed by light document, so the time is bounded by setting up the 
environment to build pandoc documents). Do you think an article like that 
will be useful? There’s some stat in ickc/travis-ci-pandoc-latex-config 
<https://github.com/ickc/travis-ci-pandoc-latex-config>.

On Wednesday, November 16, 2016 at 1:54:48 PM UTC-8, Jakob Voss wrote:

Hi Kolen,
>
> > I’m currently experimenting with Travis CI build against *all* pandoc 
> versions. The script I use:
>
> Interesting, I was also thinking how to best do this. Where can we find 
> your full script and repository?
>
> Jakob
>
​

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/5b3fea0d-48ec-4369-b41e-3588e415d2ca%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.2: Type: text/html, Size: 10563 bytes --]

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[John MacFarlane]

Why not have your script (or whatever constructs the matrix)
look at the releases page and just read off the name from
that?  All the information you need is there.

+++ Kolen Cheung [Nov 16 16 15:30 ]:
>   I see. I took your advice to change the scripting logic. I now
>   understand this is a non-issue.
>   But it seems it is still difficult to script for dual-OSes. I'm not
>   planning to do it now, but in principle I can setup a matrix that has 2
>   OS version (macOS/Linux) then lists all available pandoc versions. In
>   this case if the macOS and Linux version differs
>   (say, pandoc-1.17.1-2-amd64.deb and pandoc-1.17.1-1-osx.pkg), it seems
>   the only way to do it is either taking away the benefits of "matrix
>   expansion" (say, 2X10 builds rather than 20), or write a loop that
>   iterate the last digit until successful download.
>   On Wednesday, November 16, 2016 at 3:26:44 AM UTC-8, John MacFarlane
>   wrote:
>
>     +++ Ivan Lazar Miljenovic [Nov 16 16 12:19 ]:
>     >But why should anyone using a packaging system other than .deb have
>     >random version bumps just because someone didn't build a .deb
>     >correctly (or needs to do an update due to system libraries, etc.)?
>     >
>     >This may even be a Debian packaging requirement (I wouldn't know;
>     the
>     >closest I've had to dealing with Debian-derived distros is Ubuntu
>     for
>     >Travis-CI and Docker).
>     Yes, debian always has this versioning on the package, which
>     is independent of the upstream version.
>
>   --
>   You received this message because you are subscribed to the Google
>   Groups "pandoc-discuss" group.
>   To unsubscribe from this group and stop receiving emails from it, send
>   an email to [1]pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
>   To post to this group, send email to
>   [2]pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
>   To view this discussion on the web visit
>   [3]https://groups.google.com/d/msgid/pandoc-discuss/fe2ffb26-71af-4b01-
>   97f1-406a47e5ba6d%40googlegroups.com.
>   For more options, visit [4]https://groups.google.com/d/optout.
>
>References
>
>   1. mailto:pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org
>   2. mailto:pandoc-discuss-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org
>   3. https://groups.google.com/d/msgid/pandoc-discuss/fe2ffb26-71af-4b01-97f1-406a47e5ba6d-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org?utm_medium=email&utm_source=footer
>   4. https://groups.google.com/d/optout

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

['Jakob Voss' via pandoc-discuss]

[-- Attachment #1.1: Type: text/plain, Size: 1184 bytes --]

John MacFarlane wrote:

Why not have your script (or whatever constructs the matrix) 
> look at the releases page and just read off the name from 
> that?  All the information you need is there.
>

It's a bit more complicated but no real issue. I just created this utility 
script to download all releases and extract all pandoc binaries to use them 
in parallel. I guess it's faster than Kolen's method (which has other 
benefits) especially if you can cache the binaries folder:

https://github.com/nichtich/Pandoc-Elements/blob/cd44b41081254e5c8ad280ef39a11a8d5a8b3f22/xt/get-pandoc-releases.pl

Cheers
Jakob

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/95109c2b-d91f-4939-9150-6c5c87a6c2e7%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.2: Type: text/html, Size: 1775 bytes --]

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Kolen Cheung]

[-- Attachment #1.1: Type: text/plain, Size: 3158 bytes --]



I’m not sure caching the downloaded binaries will have any speed boost (or 
even penalty). I read probably from the travis doc saying that downloaded 
binaries shouldn’t be cached.

The reason is the cached is archived and uploaded to Amazon’s S3. And 
GitHub releases are also served from Amazon’s S3. And their server probably 
has a cache on popular packages as well. So at best the cache would only be 
as fast as downloading from Amazon’s S3, at worst, if there’s a copy of the 
pandoc version you are downloading in their server, it would be much slower 
to grab from S3’s cache.

Cache gives speed boost especially when you build from source, that those 
auxiliary files are cached rather than rebuild again. Downloading is faster 
than building, so in that case cache would be great.

Another factor might be parallel downloads, cache from S3 probably is 1 
(fast) thread only.

Depending on your need, if you need TeXLive, a tip is to use Ubuntu 12.04 
container-based, language: R combination. (Language R is community driven 
but not official). Because of the use of docker, the initializing time is 
less than 2 minutes (and boot in a few seconds). For mac however, language: 
generic and install everything yourself is faster than using language: R. 
Weird.

Their mac performance is not great. The biggest bottleneck in pandoc’s 
build time are from the 2 mac builds, and is slower than the hypothetical 
limit of 1 core (mac build) vs. 2 core (linux build). I’m interested in 
seeing how to cut down the pandoc build time. There’s a almost trivial 
trick, but sadly the time gained is negligible of the the dozens minutes.

References:

   - Caching Dependencies and Directories - Travis CI 
   <https://docs.travis-ci.com/user/caching/#How-does-the-caching-work%3F> 

On Thursday, November 17, 2016 at 1:11:37 PM UTC-8, Jakob Voss wrote:

John MacFarlane wrote:
>
> Why not have your script (or whatever constructs the matrix) 
>> look at the releases page and just read off the name from 
>> that?  All the information you need is there.
>>
>
> It's a bit more complicated but no real issue. I just created this utility 
> script to download all releases and extract all pandoc binaries to use them 
> in parallel. I guess it's faster than Kolen's method (which has other 
> benefits) especially if you can cache the binaries folder:
>
>
> https://github.com/nichtich/Pandoc-Elements/blob/cd44b41081254e5c8ad280ef39a11a8d5a8b3f22/xt/get-pandoc-releases.pl
>
> Cheers
> Jakob
>
​

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/61dc18a1-4c66-4909-8145-c90a699ccce3%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.2: Type: text/html, Size: 10022 bytes --]

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Kolen Cheung]

[-- Attachment #1.1: Type: text/plain, Size: 3498 bytes --]

This is actually a nice trick: only through this trick can one point to the 
arbitrary latest GitHub release without knowing the version!

On Thursday, November 17, 2016 at 4:42:44 AM UTC-8, John MacFarlane wrote:
>
> Why not have your script (or whatever constructs the matrix) 
> look at the releases page and just read off the name from 
> that?  All the information you need is there. 
>
> +++ Kolen Cheung [Nov 16 16 15:30 ]: 
> >   I see. I took your advice to change the scripting logic. I now 
> >   understand this is a non-issue. 
> >   But it seems it is still difficult to script for dual-OSes. I'm not 
> >   planning to do it now, but in principle I can setup a matrix that has 
> 2 
> >   OS version (macOS/Linux) then lists all available pandoc versions. In 
> >   this case if the macOS and Linux version differs 
> >   (say, pandoc-1.17.1-2-amd64.deb and pandoc-1.17.1-1-osx.pkg), it seems 
> >   the only way to do it is either taking away the benefits of "matrix 
> >   expansion" (say, 2X10 builds rather than 20), or write a loop that 
> >   iterate the last digit until successful download. 
> >   On Wednesday, November 16, 2016 at 3:26:44 AM UTC-8, John MacFarlane 
> >   wrote: 
> > 
> >     +++ Ivan Lazar Miljenovic [Nov 16 16 12:19 ]: 
> >     >But why should anyone using a packaging system other than .deb have 
> >     >random version bumps just because someone didn't build a .deb 
> >     >correctly (or needs to do an update due to system libraries, etc.)? 
> >     > 
> >     >This may even be a Debian packaging requirement (I wouldn't know; 
> >     the 
> >     >closest I've had to dealing with Debian-derived distros is Ubuntu 
> >     for 
> >     >Travis-CI and Docker). 
> >     Yes, debian always has this versioning on the package, which 
> >     is independent of the upstream version. 
> > 
> >   -- 
> >   You received this message because you are subscribed to the Google 
> >   Groups "pandoc-discuss" group. 
> >   To unsubscribe from this group and stop receiving emails from it, send 
> >   an email to [1]pandoc-discus...-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org <javascript:>. 
> >   To post to this group, send email to 
> >   [2]pandoc-...-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org <javascript:>. 
> >   To view this discussion on the web visit 
> >   [3]
> https://groups.google.com/d/msgid/pandoc-discuss/fe2ffb26-71af-4b01- 
> >   97f1-406a47e5ba6d%40googlegroups.com. 
> >   For more options, visit [4]https://groups.google.com/d/optout. 
> > 
> >References 
> > 
> >   1. mailto:pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org <javascript:> 
> >   2. mailto:pandoc-...-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org <javascript:> 
> >   3. 
> https://groups.google.com/d/msgid/pandoc-discuss/fe2ffb26-71af-4b01-97f1-406a47e5ba6d-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org?utm_medium=email&utm_source=footer 
> >   4. https://groups.google.com/d/optout 
>
>

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/110d0751-0142-42fa-8e1a-ac2093689145%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.2: Type: text/html, Size: 6993 bytes --]

Re: GitHub Pandoc Release' filename for `.deb` occasionally breaks the pattern of `-1-amd64.deb`

[Kolen Cheung]

[-- Attachment #1.1: Type: text/plain, Size: 1496 bytes --]



Looking at how travis’ community supported language R handle the pandoc 
version: travis-build/r.rb at master · travis-ci/travis-build 
<https://github.com/travis-ci/travis-build/blob/master/lib/travis/build/script/r.rb#L478-L506>, 
they actually hard coded the -1-amd64.deb pattern. And in their macOS 
build, they assume the -1 pattern isn’t there, which is sometimes not true 
(e.g. pandoc-1.17.1-1-osx.pkg).

So it seems the problem mentioned in this thread is not widely known?

I am thinking about gathering this sort of info in a ghpages, kind of like 
pandoc-extras. The current organization is in pandoc-extras: unofficial org 
for managing Pandoc extras <https://github.com/pandoc-extras>. Would @jgm 
be interested in pointing the current pandoc extras’ link in pandoc.org to 
that page? (if so, I will port the info from pandoc extras and pandoc 
filters in wiki to there).
​

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To post to this group, send email to pandoc-discuss-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/539445d0-bd80-425f-8889-f8363672442d%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.2: Type: text/html, Size: 5735 bytes --]