[Pkg-javascript-devel] Fwd: q2-composition - removed vega* from source tree

[Nilesh Patra]

Your message did not reach the js-team mailing list, hence forwarding.


-------- Forwarded Message --------
Subject: Re: q2-composition - removed vega* from source tree
Date: Fri, 25 Mar 2022 10:03:02 +0100
From: Steffen Möller <steffen_moeller at gmx.de>
To: pkg-javascript-devel at lists.debian.org
CC: Nilesh Patra <nilesh at debian.org>

[moved from the med packaging list to pkg-javascript-devel, not sure if
I am a member, will find out, please CC me, just in case]

Hi Nilesh,

On 24.03.22 22:29, Nilesh Patra wrote:
> On 25 March 2022 12:48:33 am IST, "Steffen Möller" <steffen_moeller at gmx.de> wrote:
>> I am tempted to upload the package as it is. It is apparently version 4
>> of Vega that is expected, not version 5 that just entered the archive
>> (many, many thanks!!)

Now writing to this list, I should repeat to express my gratitude for
you having addressed this. As a quick executive summary of the context:
The upstream source tree is on https://github.com/qiime2/q2-composition/
with their vega-redistribution in
https://github.com/qiime2/q2-composition/tree/master/q2_composition/assets/ancom/js
:

vega-embed.min.js
vega.min.js

and this was added four years ago.

The qiime2 packages are about identifying bacteria and yeasts that live
on us and in us and to learn about how this changes when we get sick,
when we change the diet, when we get older, ... you guess it. Also in
agriculture, it tells how the soil's bacteria change when salts change,
temperature change, pesticides are applied, ... you get the picture.

>> and they are apparently using
>> https://github.com/vega/vega-embed which we yet do not have. However,
>> those embedded vega files will affect the presentation of the results
>> but not the results themselves and as such this package is already
>> useful. I propose to craft a bug report early about the package to
>> prevent its migration to testing and talk to upstream about.
> If the package is useful, I don't see much of a problem if it migrated to testing.
It is somewhat tough to promote our qiime2 packaging to end users if
they cannot be sure that they get the full reports. So, useful, I am
confident, yes. But you lose that extra confidence in the software that
should come with using a Debian package of it. This is why I would not
want the package to migrate early but we could get it in and have the
FTPmasters check it already.
> But I agree that it makes sense to contact upstream to ask if they wouldinclude vega-embed in the monorepo as well.
>
> Or one work around (meanwhile to buy time) could be to include vega-embed inside say, debian/resources
>
> And do something like:
>
> execute_before_dh_auto_configure:
>          cp -a debian/resources $(CURDIR)/packages
I would not mind in the sense that it gives some confidence that our
package will behave exactly how upstream wants it to behave. On the
other hand, this is not how progress works. If we would find it
technically feasible to help qiime2 to update their code (does not look
like much) to become compatible with the latest version then we should
do that.
> And so this would help get vega-embed as well.
> Or in the worst case scenario, we could package node-vega-embed as a different source package.

Since Vega is used by several qiime packages now, this seems like a good
idea, except for that version problem that I interpreted from what I
read in the sources. Now, the qiime packages would likely benefit from a
transition to a later version of the vega JS, if my interpretation was
correct in the first place and I am not confused over vega-embed, and
whoever finds the time may find it rewarding to sync with upstream about it.

>> https://salsa.debian.org/med-team/q2-composition
>>
>> What do you think?
> Let me know what you think, and also, this conversation should not be happening in the -med@ mailing list.
>
> Please consider sending query about this to the pkg-javascript-devel mailing list instead, since there are people who are specifically involved with JS stuff; even the vega.js maintainer would be easy to reach out.

"Hello!" - and many thanks again

Steffen