(Forwarding message from Discourse here, but better keep the discussion there: https://pandas.discourse.group/t/remove-old-mailing-lists/) I don't think we should keep the existing mailing lists for too long in parallel with the new Discourse. Otherwise we'll keep the two systems, and have a mess with the discussions. My proposal is: - Announce Discourse via twitter and the pydata mailing list - Leave couple of weeks to see the there is no unexpected issue with Discourse - On the 25th of November, delete the previous mailing lists (pandas-dev, and the core team google group), and keep just Discourse I guess the pydata google group has a broader scope than pandas, and should be left as is for now. Any objection?
Here’s a (possible) guide to make discourse work like a mailing list . I haven’t tried it yet but am hoping to soon: https://discourse.mozilla.org/t/how-do-i-use-discourse-via-email/15279 On Mon, Nov 11, 2019 at 4:46 PM Marc Garcia <garcia.marc@gmail.com> wrote:
(Forwarding message from Discourse here, but better keep the discussion there: https://pandas.discourse.group/t/remove-old-mailing-lists/)
I don't think we should keep the existing mailing lists for too long in parallel with the new Discourse. Otherwise we'll keep the two systems, and have a mess with the discussions.
My proposal is: - Announce Discourse via twitter and the pydata mailing list - Leave couple of weeks to see the there is no unexpected issue with Discourse - On the 25th of November, delete the previous mailing lists (pandas-dev, and the core team google group), and keep just Discourse
I guess the pydata google group has a broader scope than pandas, and should be left as is for now.
Any objection? _______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
To be honest, I totally missed this, having thought that discourse was an experiment in community envolvement and not an immediate replacement for lists, even less so for devs lists. Precisely because I missed this, I totally accept whatever other devs, who spent time planning this, think works best. But the more the system works similarly as mailing lists (see Jeffrey's mail), the better I like it. Cheers, Pietro Il giorno lun, 11/11/2019 alle 18.47 -0800, Jeffrey Tratner ha scritto:
Here’s a (possible) guide to make discourse work like a mailing list . I haven’t tried it yet but am hoping to soon: https://discourse.mozilla.org/t/how-do-i-use-discourse-via-email/15279
On Mon, Nov 11, 2019 at 4:46 PM Marc Garcia <garcia.marc@gmail.com> wrote:
(Forwarding message from Discourse here, but better keep the discussion there: https://pandas.discourse.group/t/remove-old-mailing-lists/)
I don't think we should keep the existing mailing lists for too long in parallel with the new Discourse. Otherwise we'll keep the two systems, and have a mess with the discussions.
My proposal is: - Announce Discourse via twitter and the pydata mailing list - Leave couple of weeks to see the there is no unexpected issue with Discourse - On the 25th of November, delete the previous mailing lists (pandas-dev, and the core team google group), and keep just Discourse
I guess the pydata google group has a broader scope than pandas, and should be left as is for now.
Any objection? _______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
Let's actually keep the discussion here, to get feedback from people who haven't signed up for discourse. I agree that we don't want to run these in parallel. So if we're going to make the switch we'll want to shut this list down and at least announce it on the PyData google group. I don't know what mailman supports, but we won't want to really delete the pandas-dev mailing list. https://mail.python.org/pipermail/pandas-dev/ is a valuable archive of past discussions. Hopefully we're able to put this list in archive mode. We can discuss this on Wednesday's call. On Tue, Nov 12, 2019 at 2:32 AM Pietro Battiston <me@pietrobattiston.it> wrote:
To be honest, I totally missed this, having thought that discourse was an experiment in community envolvement and not an immediate replacement for lists, even less so for devs lists.
Precisely because I missed this, I totally accept whatever other devs, who spent time planning this, think works best.
But the more the system works similarly as mailing lists (see Jeffrey's mail), the better I like it.
Cheers,
Pietro
Il giorno lun, 11/11/2019 alle 18.47 -0800, Jeffrey Tratner ha scritto:
Here’s a (possible) guide to make discourse work like a mailing list . I haven’t tried it yet but am hoping to soon: https://discourse.mozilla.org/t/how-do-i-use-discourse-via-email/15279
On Mon, Nov 11, 2019 at 4:46 PM Marc Garcia <garcia.marc@gmail.com> wrote:
(Forwarding message from Discourse here, but better keep the discussion there: https://pandas.discourse.group/t/remove-old-mailing-lists/)
I don't think we should keep the existing mailing lists for too long in parallel with the new Discourse. Otherwise we'll keep the two systems, and have a mess with the discussions.
My proposal is: - Announce Discourse via twitter and the pydata mailing list - Leave couple of weeks to see the there is no unexpected issue with Discourse - On the 25th of November, delete the previous mailing lists (pandas-dev, and the core team google group), and keep just Discourse
I guess the pydata google group has a broader scope than pandas, and should be left as is for now.
Any objection? _______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
I'm not very active in pandas development anymore, but for what it's worth, I find mailing lists much easier to follow casually. Everyone once in a while I see something interesting and chime in. I would be unlikely to do that on Discourse. On Tue, Nov 12, 2019 at 3:47 AM Tom Augspurger <tom.augspurger88@gmail.com> wrote:
Let's actually keep the discussion here, to get feedback from people who haven't signed up for discourse.
I agree that we don't want to run these in parallel. So if we're going to make the switch we'll want to shut this list down and at least announce it on the PyData google group.
I don't know what mailman supports, but we won't want to really delete the pandas-dev mailing list. https://mail.python.org/pipermail/pandas-dev/ is a valuable archive of past discussions. Hopefully we're able to put this list in archive mode.
We can discuss this on Wednesday's call.
On Tue, Nov 12, 2019 at 2:32 AM Pietro Battiston <me@pietrobattiston.it> wrote:
To be honest, I totally missed this, having thought that discourse was an experiment in community envolvement and not an immediate replacement for lists, even less so for devs lists.
Precisely because I missed this, I totally accept whatever other devs, who spent time planning this, think works best.
But the more the system works similarly as mailing lists (see Jeffrey's mail), the better I like it.
Cheers,
Pietro
Il giorno lun, 11/11/2019 alle 18.47 -0800, Jeffrey Tratner ha scritto:
Here’s a (possible) guide to make discourse work like a mailing list . I haven’t tried it yet but am hoping to soon: https://discourse.mozilla.org/t/how-do-i-use-discourse-via-email/15279
On Mon, Nov 11, 2019 at 4:46 PM Marc Garcia <garcia.marc@gmail.com> wrote:
(Forwarding message from Discourse here, but better keep the discussion there: https://pandas.discourse.group/t/remove-old-mailing-lists/)
I don't think we should keep the existing mailing lists for too long in parallel with the new Discourse. Otherwise we'll keep the two systems, and have a mess with the discussions.
My proposal is: - Announce Discourse via twitter and the pydata mailing list - Leave couple of weeks to see the there is no unexpected issue with Discourse - On the 25th of November, delete the previous mailing lists (pandas-dev, and the core team google group), and keep just Discourse
I guess the pydata google group has a broader scope than pandas, and should be left as is for now.
Any objection? _______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
I'm in the same position as Stephan. I would be interested if there was some broader PyData wide discourse though. I rarely interact here though, so I wouldn't optimize for me. On Tue, Nov 12, 2019 at 10:34 AM Stephan Hoyer <shoyer@gmail.com> wrote:
I'm not very active in pandas development anymore, but for what it's worth, I find mailing lists much easier to follow casually. Everyone once in a while I see something interesting and chime in. I would be unlikely to do that on Discourse.
On Tue, Nov 12, 2019 at 3:47 AM Tom Augspurger <tom.augspurger88@gmail.com> wrote:
Let's actually keep the discussion here, to get feedback from people who haven't signed up for discourse.
I agree that we don't want to run these in parallel. So if we're going to make the switch we'll want to shut this list down and at least announce it on the PyData google group.
I don't know what mailman supports, but we won't want to really delete the pandas-dev mailing list. https://mail.python.org/pipermail/pandas-dev/ is a valuable archive of past discussions. Hopefully we're able to put this list in archive mode.
We can discuss this on Wednesday's call.
On Tue, Nov 12, 2019 at 2:32 AM Pietro Battiston <me@pietrobattiston.it> wrote:
To be honest, I totally missed this, having thought that discourse was an experiment in community envolvement and not an immediate replacement for lists, even less so for devs lists.
Precisely because I missed this, I totally accept whatever other devs, who spent time planning this, think works best.
But the more the system works similarly as mailing lists (see Jeffrey's mail), the better I like it.
Cheers,
Pietro
Il giorno lun, 11/11/2019 alle 18.47 -0800, Jeffrey Tratner ha scritto:
Here’s a (possible) guide to make discourse work like a mailing list . I haven’t tried it yet but am hoping to soon: https://discourse.mozilla.org/t/how-do-i-use-discourse-via-email/15279
On Mon, Nov 11, 2019 at 4:46 PM Marc Garcia <garcia.marc@gmail.com> wrote:
(Forwarding message from Discourse here, but better keep the discussion there: https://pandas.discourse.group/t/remove-old-mailing-lists/)
I don't think we should keep the existing mailing lists for too long in parallel with the new Discourse. Otherwise we'll keep the two systems, and have a mess with the discussions.
My proposal is: - Announce Discourse via twitter and the pydata mailing list - Leave couple of weeks to see the there is no unexpected issue with Discourse - On the 25th of November, delete the previous mailing lists (pandas-dev, and the core team google group), and keep just Discourse
I guess the pydata google group has a broader scope than pandas, and should be left as is for now.
Any objection? _______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
There was a separate thread and a GitHub issue to discuss possible options, including the ones we currently have, and a PyData wide Discourse. I haven't used Discourse before, so can't really tell much about how it compares. But most projects of the ecosystem have moved or are planning to moved to Discourse, and that was the preferred option after the previous discussions. As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address. You can check those past conversations, and if there is a new reason to use something different than Discourse, please let us know asap. Otherwise, please sign up to Discourse at pandas.discourse.group, and let me know if you have any objection to the proposed plan to make the current lists inactive in couple of weeks (keeping the archive accessible as Tom said). On Tue, 12 Nov 2019, 19:51 Matthew Rocklin, <mrocklin@gmail.com> wrote:
I'm in the same position as Stephan.
I would be interested if there was some broader PyData wide discourse though.
I rarely interact here though, so I wouldn't optimize for me.
On Tue, Nov 12, 2019 at 10:34 AM Stephan Hoyer <shoyer@gmail.com> wrote:
I'm not very active in pandas development anymore, but for what it's worth, I find mailing lists much easier to follow casually. Everyone once in a while I see something interesting and chime in. I would be unlikely to do that on Discourse.
On Tue, Nov 12, 2019 at 3:47 AM Tom Augspurger < tom.augspurger88@gmail.com> wrote:
Let's actually keep the discussion here, to get feedback from people who haven't signed up for discourse.
I agree that we don't want to run these in parallel. So if we're going to make the switch we'll want to shut this list down and at least announce it on the PyData google group.
I don't know what mailman supports, but we won't want to really delete the pandas-dev mailing list. https://mail.python.org/pipermail/pandas-dev/ is a valuable archive of past discussions. Hopefully we're able to put this list in archive mode.
We can discuss this on Wednesday's call.
On Tue, Nov 12, 2019 at 2:32 AM Pietro Battiston <me@pietrobattiston.it> wrote:
To be honest, I totally missed this, having thought that discourse was an experiment in community envolvement and not an immediate replacement for lists, even less so for devs lists.
Precisely because I missed this, I totally accept whatever other devs, who spent time planning this, think works best.
But the more the system works similarly as mailing lists (see Jeffrey's mail), the better I like it.
Cheers,
Pietro
Il giorno lun, 11/11/2019 alle 18.47 -0800, Jeffrey Tratner ha scritto:
Here’s a (possible) guide to make discourse work like a mailing list . I haven’t tried it yet but am hoping to soon:
https://discourse.mozilla.org/t/how-do-i-use-discourse-via-email/15279
On Mon, Nov 11, 2019 at 4:46 PM Marc Garcia <garcia.marc@gmail.com> wrote:
(Forwarding message from Discourse here, but better keep the discussion there: https://pandas.discourse.group/t/remove-old-mailing-lists/)
I don't think we should keep the existing mailing lists for too long in parallel with the new Discourse. Otherwise we'll keep the two systems, and have a mess with the discussions.
My proposal is: - Announce Discourse via twitter and the pydata mailing list - Leave couple of weeks to see the there is no unexpected issue with Discourse - On the 25th of November, delete the previous mailing lists (pandas-dev, and the core team google group), and keep just Discourse
I guess the pydata google group has a broader scope than pandas, and should be left as is for now.
Any objection? _______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
Marc, is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ? Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs. Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light? Thanks, Pietro
That's the GitHub issue, most of the discussion was in a thread in this (pandas-dev) list. I'm in my phone and can't look for the thread now, but I think it started in the conversation about the website hosting, and then Joris created a separate thread specific to Discourse. As I said, I have no preference on Discourse (never used it as I said), but I think what we have now is suboptimal, and would be great to have something better. But we had that discussion several weeks ago. We decided to move forward with Discourse at that time. I spent many hours learning about it, and setting it up. So, I'm not really looking forward to start again with this. I'm happy to hand over this to you, if you want to research further and lead the discussion. And implement whatever is best. But if I need to continue spending time on this myself, I'd appreciate if you can find solutions and not problems. I have no idea about how to set up Discourse as a mailing list, or how to do it for subcommunities. But if you have a specific way you want it to work, please do the research, and propose (and implement) the best solution for all us. Or is your proposal to stay with what we have. Thanks! On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it> wrote:
Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it> wrote:
Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
For what it’s worth - all the discourse channels have an RSS feed, so using your preferred RSS reader makes the occasional dipping in and out straight forward. For intermittent readers, It also offers the benefit that you easily cast your eye over all that has transpired since you last dipped in. cheers
On 13 Nov 2019, at 7:12 am, Marc Garcia <garcia.marc@gmail.com> wrote:
That's the GitHub issue, most of the discussion was in a thread in this (pandas-dev) list. I'm in my phone and can't look for the thread now, but I think it started in the conversation about the website hosting, and then Joris created a separate thread specific to Discourse.
As I said, I have no preference on Discourse (never used it as I said), but I think what we have now is suboptimal, and would be great to have something better. But we had that discussion several weeks ago. We decided to move forward with Discourse at that time. I spent many hours learning about it, and setting it up. So, I'm not really looking forward to start again with this. I'm happy to hand over this to you, if you want to research further and lead the discussion. And implement whatever is best.
But if I need to continue spending time on this myself, I'd appreciate if you can find solutions and not problems. I have no idea about how to set up Discourse as a mailing list, or how to do it for subcommunities. But if you have a specific way you want it to work, please do the research, and propose (and implement) the best solution for all us. Or is your proposal to stay with what we have.
Thanks!
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it <mailto:me@pietrobattiston.it>> wrote: Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 <https://github.com/pandas-dev/pandas/issues/27903> ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it <mailto:me@pietrobattiston.it>> wrote: Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 <https://github.com/pandas-dev/pandas/issues/27903> ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
I was granting admin permissions to maintainers on the new Discourse, and looks like we've got a limit of 5. Spoke with Discourse support, and they want us to pay >$300 per month to have more. I think we could live with this, but I'm a bit worried about their open source plan being more of a freemium plan they use to make big open source projects pay. I was reading the details of the conditions, and seems like we also have bandwidth limits we may reach. An alternative is to host the instance ourselves. If we do that, we can also consider other alternatives. Just seem Flarum, which is in beta, but at a first glance it looks like it manages subcategories in a way that would make it possible to have a NumFOCUS broad forum. Not sure if they have the feature of letting categories/subcategories behave as mailing lists, need to research. What are your thoughts? On Tue, Nov 12, 2019 at 9:12 PM Marc Garcia <garcia.marc@gmail.com> wrote:
That's the GitHub issue, most of the discussion was in a thread in this (pandas-dev) list. I'm in my phone and can't look for the thread now, but I think it started in the conversation about the website hosting, and then Joris created a separate thread specific to Discourse.
As I said, I have no preference on Discourse (never used it as I said), but I think what we have now is suboptimal, and would be great to have something better. But we had that discussion several weeks ago. We decided to move forward with Discourse at that time. I spent many hours learning about it, and setting it up. So, I'm not really looking forward to start again with this. I'm happy to hand over this to you, if you want to research further and lead the discussion. And implement whatever is best.
But if I need to continue spending time on this myself, I'd appreciate if you can find solutions and not problems. I have no idea about how to set up Discourse as a mailing list, or how to do it for subcommunities. But if you have a specific way you want it to work, please do the research, and propose (and implement) the best solution for all us. Or is your proposal to stay with what we have.
Thanks!
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it> wrote:
Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it> wrote:
Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
Marc, on whether Discourse (or Flarum) is the right tool to best reach/host our (pydata) community, while not being a fan of Discourse from my very limited experience (GNOME), I was not rhetorical when I wrote that I trust what you, and other people who spent time on this, think. I say this not just because I don't have time now to devote to the issue, but also because you already did great things for pandas in the past in terms of community involvement. You definitely know how to reach people, better than me. My previous two emails were just pointing out that while the pydata ML and gitter were probably a suboptimal way to involve our community, the devs MLs seemed to me to just work fine, and I didn't read of specific complaints that would justify closing them. Sure, some people had missed the fact that pandas-dev is open to non-core contributors, but I think we did some steps in clarifying the options to get in contact with us: https://github.com/pandas-dev/pandas-website/issues/68 Then, I might have missed some arguments/conclusions: in this case, please let me know, and I will stop bothering. But I'm not "against Discourse". Cheers, Pietro Il giorno mer, 13/11/2019 alle 03.10 +0100, Marc Garcia ha scritto:
I was granting admin permissions to maintainers on the new Discourse, and looks like we've got a limit of 5. Spoke with Discourse support, and they want us to pay >$300 per month to have more. I think we could live with this, but I'm a bit worried about their open source plan being more of a freemium plan they use to make big open source projects pay. I was reading the details of the conditions, and seems like we also have bandwidth limits we may reach.
An alternative is to host the instance ourselves. If we do that, we can also consider other alternatives. Just seem Flarum, which is in beta, but at a first glance it looks like it manages subcategories in a way that would make it possible to have a NumFOCUS broad forum. Not sure if they have the feature of letting categories/subcategories behave as mailing lists, need to research.
What are your thoughts?
On Tue, Nov 12, 2019 at 9:12 PM Marc Garcia <garcia.marc@gmail.com> wrote:
That's the GitHub issue, most of the discussion was in a thread in this (pandas-dev) list. I'm in my phone and can't look for the thread now, but I think it started in the conversation about the website hosting, and then Joris created a separate thread specific to Discourse.
As I said, I have no preference on Discourse (never used it as I said), but I think what we have now is suboptimal, and would be great to have something better. But we had that discussion several weeks ago. We decided to move forward with Discourse at that time. I spent many hours learning about it, and setting it up. So, I'm not really looking forward to start again with this. I'm happy to hand over this to you, if you want to research further and lead the discussion. And implement whatever is best.
But if I need to continue spending time on this myself, I'd appreciate if you can find solutions and not problems. I have no idea about how to set up Discourse as a mailing list, or how to do it for subcommunities. But if you have a specific way you want it to work, please do the research, and propose (and implement) the best solution for all us. Or is your proposal to stay with what we have.
Thanks!
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it
wrote: Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it
wrote: Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
We can probably continue in the call, it'll be easier. My only point was that the discussion on whether we want Discourse or not already happened. And we also discussed on being able to use Discourse as a mailing list. So, if anybody wants to reopen that discussion, it's worth to catch up with those discussions, and propose specific changes to the current plan. We spent a significant amount of hours on this, and I don't think general opinions or preferences are helpful at this point. It's very difficult to move forward if we keep rediscussing things without proposing alternatives to what was previously agreed. I see your point now, but even if pandas-dev is working well, and there is no reason to close it, there is no reason to continue using it if Discourse can do exactly the same, and more things. In my opinion, people who have a preference on using mailing lists (and that may be me too) shouldn't be a reason to not move forward. The change will be transparent and you'll just have to sign up in Discourse, and start using a different email address to send messages. And other people will benefit from the new features that those platforms offer. And things will be easier to manage and easier to understand by newcomers, if we have a single platform for the communications. Does this make sense? Or is there something I'm missing? On Wed, 13 Nov 2019, 09:21 Pietro Battiston, <me@pietrobattiston.it> wrote:
Marc,
on whether Discourse (or Flarum) is the right tool to best reach/host our (pydata) community, while not being a fan of Discourse from my very limited experience (GNOME), I was not rhetorical when I wrote that I trust what you, and other people who spent time on this, think. I say this not just because I don't have time now to devote to the issue, but also because you already did great things for pandas in the past in terms of community involvement. You definitely know how to reach people, better than me.
My previous two emails were just pointing out that while the pydata ML and gitter were probably a suboptimal way to involve our community, the devs MLs seemed to me to just work fine, and I didn't read of specific complaints that would justify closing them. Sure, some people had missed the fact that pandas-dev is open to non-core contributors, but I think we did some steps in clarifying the options to get in contact with us: https://github.com/pandas-dev/pandas-website/issues/68
Then, I might have missed some arguments/conclusions: in this case, please let me know, and I will stop bothering. But I'm not "against Discourse".
Cheers,
Pietro
Il giorno mer, 13/11/2019 alle 03.10 +0100, Marc Garcia ha scritto:
I was granting admin permissions to maintainers on the new Discourse, and looks like we've got a limit of 5. Spoke with Discourse support, and they want us to pay >$300 per month to have more. I think we could live with this, but I'm a bit worried about their open source plan being more of a freemium plan they use to make big open source projects pay. I was reading the details of the conditions, and seems like we also have bandwidth limits we may reach.
An alternative is to host the instance ourselves. If we do that, we can also consider other alternatives. Just seem Flarum, which is in beta, but at a first glance it looks like it manages subcategories in a way that would make it possible to have a NumFOCUS broad forum. Not sure if they have the feature of letting categories/subcategories behave as mailing lists, need to research.
What are your thoughts?
On Tue, Nov 12, 2019 at 9:12 PM Marc Garcia <garcia.marc@gmail.com> wrote:
That's the GitHub issue, most of the discussion was in a thread in this (pandas-dev) list. I'm in my phone and can't look for the thread now, but I think it started in the conversation about the website hosting, and then Joris created a separate thread specific to Discourse.
As I said, I have no preference on Discourse (never used it as I said), but I think what we have now is suboptimal, and would be great to have something better. But we had that discussion several weeks ago. We decided to move forward with Discourse at that time. I spent many hours learning about it, and setting it up. So, I'm not really looking forward to start again with this. I'm happy to hand over this to you, if you want to research further and lead the discussion. And implement whatever is best.
But if I need to continue spending time on this myself, I'd appreciate if you can find solutions and not problems. I have no idea about how to set up Discourse as a mailing list, or how to do it for subcommunities. But if you have a specific way you want it to work, please do the research, and propose (and implement) the best solution for all us. Or is your proposal to stay with what we have.
Thanks!
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it
wrote: Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it
wrote: Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
On Wed, 13 Nov 2019 at 12:08, Marc Garcia <garcia.marc@gmail.com> wrote:
We can probably continue in the call, it'll be easier.
My only point was that the discussion on whether we want Discourse or not already happened.
There was indeed a mailing thread, but I don't think there was much discussion (and I am not blaming you for that, to be clear ;)) or involvement of many people. For such a change of moving away from the current mailing lists, personally I think we should have more buy in from the people actively using those mailing lists (also for our internal one). So I would rather welcome that we now finally have some discussion about it. And IMO we should also not defer everything to a call. Not all people are present then, mailing list (or discourse :)) discussions about significant topics are important as well. That said: I am in favor of trying out discourse (although the limitations of their free plan might sound problematic. It might be worth checking with other projects using it (jupyter, matplotlib) how they are dealing with that, or are they hosting themselves?). But I think it should be a "trying out", and not already saying "in a month we will close the mailing lists". So I am in favor of trying out discourse for a while, so we can then evaluate if it works and if we want to archive or discourage the mailing lists. Regarding the usage as a mailing list: there is also a "mailing list mode" in the preferences (not fully sure what the difference is with the previously mentioned link about configuring it to use with email) Joris
And we also discussed on being able to use Discourse as a mailing list. So, if anybody wants to reopen that discussion, it's worth to catch up with those discussions, and propose specific changes to the current plan. We spent a significant amount of hours on this, and I don't think general opinions or preferences are helpful at this point. It's very difficult to move forward if we keep rediscussing things without proposing alternatives to what was previously agreed.
I see your point now, but even if pandas-dev is working well, and there is no reason to close it, there is no reason to continue using it if Discourse can do exactly the same, and more things.
In my opinion, people who have a preference on using mailing lists (and that may be me too) shouldn't be a reason to not move forward. The change will be transparent and you'll just have to sign up in Discourse, and start using a different email address to send messages. And other people will benefit from the new features that those platforms offer. And things will be easier to manage and easier to understand by newcomers, if we have a single platform for the communications.
Does this make sense? Or is there something I'm missing?
On Wed, 13 Nov 2019, 09:21 Pietro Battiston, <me@pietrobattiston.it> wrote:
Marc,
on whether Discourse (or Flarum) is the right tool to best reach/host our (pydata) community, while not being a fan of Discourse from my very limited experience (GNOME), I was not rhetorical when I wrote that I trust what you, and other people who spent time on this, think. I say this not just because I don't have time now to devote to the issue, but also because you already did great things for pandas in the past in terms of community involvement. You definitely know how to reach people, better than me.
My previous two emails were just pointing out that while the pydata ML and gitter were probably a suboptimal way to involve our community, the devs MLs seemed to me to just work fine, and I didn't read of specific complaints that would justify closing them. Sure, some people had missed the fact that pandas-dev is open to non-core contributors, but I think we did some steps in clarifying the options to get in contact with us: https://github.com/pandas-dev/pandas-website/issues/68
Then, I might have missed some arguments/conclusions: in this case, please let me know, and I will stop bothering. But I'm not "against Discourse".
Cheers,
Pietro
Il giorno mer, 13/11/2019 alle 03.10 +0100, Marc Garcia ha scritto:
I was granting admin permissions to maintainers on the new Discourse, and looks like we've got a limit of 5. Spoke with Discourse support, and they want us to pay >$300 per month to have more. I think we could live with this, but I'm a bit worried about their open source plan being more of a freemium plan they use to make big open source projects pay. I was reading the details of the conditions, and seems like we also have bandwidth limits we may reach.
An alternative is to host the instance ourselves. If we do that, we can also consider other alternatives. Just seem Flarum, which is in beta, but at a first glance it looks like it manages subcategories in a way that would make it possible to have a NumFOCUS broad forum. Not sure if they have the feature of letting categories/subcategories behave as mailing lists, need to research.
What are your thoughts?
On Tue, Nov 12, 2019 at 9:12 PM Marc Garcia <garcia.marc@gmail.com> wrote:
That's the GitHub issue, most of the discussion was in a thread in this (pandas-dev) list. I'm in my phone and can't look for the thread now, but I think it started in the conversation about the website hosting, and then Joris created a separate thread specific to Discourse.
As I said, I have no preference on Discourse (never used it as I said), but I think what we have now is suboptimal, and would be great to have something better. But we had that discussion several weeks ago. We decided to move forward with Discourse at that time. I spent many hours learning about it, and setting it up. So, I'm not really looking forward to start again with this. I'm happy to hand over this to you, if you want to research further and lead the discussion. And implement whatever is best.
But if I need to continue spending time on this myself, I'd appreciate if you can find solutions and not problems. I have no idea about how to set up Discourse as a mailing list, or how to do it for subcommunities. But if you have a specific way you want it to work, please do the research, and propose (and implement) the best solution for all us. Or is your proposal to stay with what we have.
Thanks!
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it
wrote: Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
On Tue, 12 Nov 2019, 20:46 Pietro Battiston, <me@pietrobattiston.it
wrote: Marc,
is the thread you refer to https://github.com/pandas-dev/pandas/issues/27903 ?
Again, I didn't participate in the discussion, I am not contributing much lately, and so my opinion is not very important. But I had followed the discussion, I now checked the thread above, and while I derived the idea that Discourse could (and probably should) replace the pydata ML as a frontend to users (ideally of pydata, not just pandas), I fail to see a real discussion, or even just arguments, in favour of replacing the devs MLs.
Il giorno mar, 12/11/2019 alle 20.21 +0100, Marc Garcia ha scritto:
[...]
As it has been said, Discourse can be set up to work as a mailing list for people who don't want to use the Discourse interface. So, I don't think this should be a reason to not move forward, you'll just need to subscribe elsewhere, and send messages to a different email address.
I just read the FAQ about how to "use Discourse via email"... I guess we can live with filtering incoming emails by List-ID, but I don't understand whether there will be multiple emails address to write to, for the different "subcommunities". Could you shed some light?
Thanks,
Pietro
_______________________________________________ Pandas-dev mailing list Pandas-dev@python.org https://mail.python.org/mailman/listinfo/pandas-dev
participants (8)
-
Christopher Short -
Jeffrey Tratner -
Joris Van den Bossche -
Marc Garcia -
Matthew Rocklin -
Pietro Battiston -
Stephan Hoyer -
Tom Augspurger