Debian's patches #23
Comments
|
Hi @paravoid
You're welcome. I'm slowly gathering people want Mutt to evolve.
Two weeks ago, I wrote to Antonio Radici, Christoph Berg (maintainers of the mutt-patched package). No reply. It's good to hear from a Debian developer.
That makes me angry. (With Mutt, not Debian).
Fabian Groffen (@grobian) has the same situation with Gentoo.
mutt-kz branched off from Mutt years ago and was starting to drift away. I pull all of Karel Zak's (@karelzak) changes into NeoMutt
Thanks for the breakdown. Very helpful.
Great. Let me know their requirements.
Yes. I'll get to work.
OK, here's the status of [devel/*]
I shall have a look at them. "Bringing together all the Mutt code" -- For now, I need to be a "patch curator".
That would be nice. If it's helpful, I'm happy to add distro-specific dirs to the feature branches. e.g.
(and give write privs to the devs to keep things up-to-date).
As I'm finding with Gentoo, there'll always be some conflicts.
Absolutely. Rich / FlatCap |
|
Which versions of Mutt are Debian's patches for?
|
|
Hey @flatcap, I am one of the current maintainers of mutt in Debian/Ubuntu (@ChristophBerg) did forward me your mail, but I forgot to reply, sorry! The patches @paravoid linked to are currently against mutt 1.6.0. @mfvescovi wanted to update them to 1.6.1, but I don't think he had the time yet. I think syncing mutt-patched and neomutt would be a good idea. Or dropping it altogether as mutt itself gains all features... Wishful thinking :) Anyways, My todo says to try your sidebar instead of ours in the next days, maybe even more. Thanks for all the work you did! |
No worries :-)
Good. I'll be a happy man when there are .deb's with the new sidebar.
Kevin's new window patches, for Mutt, show he's keen to integrate Sidebar at some point. Fabian is keen to cherry-pick/transplant default-branch code into Gentoo, if it gets him cool features. If there's demand, I might consider maintaining a set of patches against default.
Note: Some config has changed since the 2015-11-11 patch. Any problems, or questions, let me know.
You're welcome. Lots still to do. |
|
OK, I had another look. In terms of immediate next steps, to allow us to reuse at least a subset of neomutt's patches for a first stab without any regressions (ifdef, trash, sidebar, nntp):
The timeout-hook and multiple-fcc patches would be next in line, followed by the rest of Debian's custom patches — but all these can wait a little bit I think. |
|
@paravoid
They compile. They appear to work. |
|
A few fixes here and there and three new patch branches:
The patches look good and everything compiles. |
|
I'd just like to say "weeeeeeeeeeeeeeeee" and shall lurk on this issue from now on. Thanks to all involved :) |
Save the celebrations for later, we've got work to do :-)
Any Deb-dev knowledge you can contribute will be welcome. |
|
@flatcap, thank you so much for doing this integration work for us. I pushed a wip-neomutt branch into pkg-mutt, partially based on your work. We discussed it again within the Debian maintainers, and we are considering instead to switch to the "pristine" neomutt, with the entirety of the patches and in your order, to replace mutt-patched (and possibly mutt) entirely with it. What would help us tremendously, I think, would be what I mentioned in my last comment above: get rid of all of our upstream patches in favor of upstream ones, especially compressed folders and NNTP (that are halfway there :)). The rest of our upstream patches can follow after that, assuming you have the time and will to review them and incorporate them :) By the way, a question that inevitably came up while we were discussing our plans was: where does vanilla mutt stands in all this? We know they have been traditionally slow/reluctant to respond, but some work seems to be happening upstream now (including the windows stuff you already mentioned), so I'm wondering if we can expect some movement there as well, especially considering the curation work that you have been doing. |
Great stuff.
Message understood :-) I shall make some time this weekend and sort them out.
I've got a plan for that (open to discussion). Each distro can have their own bug-fix/tweak branch. Anything that's likely to be universally wanted, I'll put into NeoMutt's bug-fix branch and then push upstream to Mutt.
Good question. Most code pushed to Mutt will go into the default branch. These changes won't be public (merge into the stable branch) until everything in default is stable. That's likely to be many months. But, that probably doesn't matter too much. Each bug that's accepted will make our jobs easier (just not immediately). Git's perfectly happy to have the same commit in multiple places and it'll do the right thing when everything gets merged. |
|
just tested the wip branch from @paravoid and loving it so far. I'd totally make mutt-patched == neomutt in the future and move the non-bugfix patches out of mutt to mutt-patched :) |
|
@RichiH don't have my debian gear with me, but I think @mfvescovi wanted to do an experimental upload soon. |
|
On 2016-05-17 at 17:20 (CEST), Evgeni Golov wrote:
@RichiH, done. Enjoy! |
|
I have been using neomutt for a few days now, and there's only one issue I am away of: #29 Note that this might be existent in older versions as well, I never really used sidebar before. |
|
As per #29 an updated build might make sense, else I will just wait for the next release to test again. |
|
Wouldn't it make more sense to keep Debian's mutt package as it is and to create a seperate neomutt package? I am afraid, that the pristine mutt will be overpatched and at some point gets non maintainable? I was maintaining mutt-ng in Debian some years ago and the package coexists well to the pristine mutt one. If, well maybe in some decades ;-), pristine mutt will integrate the neomutt patches we can merge the two Debian packages. I can package and maintain neomutt as a seperate Debian package with a replacement to the pristine mutt/mutt-patched package so that the user can decide which one to use. From the mainataining point of view this would make more sense? |
|
Hi @riesebie, welcome to the discussion.
First, it would need to be updated to the latest versions of the patches. They contain bug-fixes and other improvements, like documentation. I don't want to get bug reports for things I can't do anything about.
It's a possibility, but... Most of the patches, out there, are already integrated, so there's not much more to add. Historically, the Mutt devs have been too conservative to new ideas and this has pushed away lots of developers. I want to encourage people to think of new ideas. I'd be happy to see a NeoMutt-devel branch, full of dangerous ideas!
Hopefully that won't happen. We're all volunteers and little time to waste. That's why I've been keen to bring the distros together. It's something the Mutt community should have done years ago. If you haven't already read it, see my Patch Tree Proposal. The distros can decide how much they want to include (to a degree). If others want to make their own patch sets, that's fine, but I don't have time to help them.
Thanks for the offer, but if @mfvescovi @evgeni are happy with a full NeoMutt, then there's no need.
From my point of view, giving everyone the same thing is much easier :-) Thanks for your ideas. Keep them coming. |
|
@riesebie I've thought of that too (both the danger of having a significant patchset to maintain forward and the possible need for a "vanilla" mutt). The problem is of course that the vanilla mutt package has accumulated quite a few patches already, several of which are invasive but are included in neomutt already (ifdef, trash, compressed folders, etc.) — and that's even without the mutt-patched variant that has a bunch more. I'm new to this and not very familiar with mutt-ng's history. Could you perhaps elaborate on what happened with it so that we can learn from history and not repeat the same mistakes? @flatcap seems to be doing a great job driving neomutt forward, but also looping in all the downstreams as well, so I'm hopeful this effort will stick even if he disappears for whatever reason. We already have to port forward our large upstream featuresets across mutt's releases, so it seems worthwhile to benefit from each other's work between the distributions and commit into a single repository. (that may be a little naive, I suppose :) In general, my opinion tends to lean towards dropping the original mutt upstream in favor of neomutt and merge as many debian/patches as we can in this new upstream. This is a LibreOffice/OpenOffice.org or glibc/eglibc deal (or gcc/egcs if you want to go farther out), as far as I can tell. On that note and to get back to on-topic for neomutt: @flatcap, thanks you so much for working on compressed and NNTP and graduating them to stable, that's awesome to hear :) In terms of user-visiable features, this leaves:
|
Er... That sounds like you're planning a coup :-)
There's more testing and tidying I want to do, but it's good enough.
It sort-of works, but it should take into account the current sort order. |
|
Just note... I prefer to use NeoMutt than directly mutt-kz. I'll keep mutt-kz as git tree, but for end-users and downstream distribution is better to have all integrated to the one package. It would be also nice to have the same [Neo]Mutt in all distributions to minimize difference and distro-specific patches. We need to share things... |
|
/me nods |
Thanks for the vote of confidence :-)
I'm happy with that, but I didn't want to presume. You've worked hard and got a lot of followers. I didn't want to undermine that.
You may not have read my Patch Tree Proposal and see my Patch Sets. Progress is being made. |
|
With the recent 1.6.1-2 Debian upload, we've come much closer to neomutt (this release is actually neomutt). See https://anonscm.debian.org/cgit/pkg-mutt/mutt.git/log/?h=experimental for the full changelog and https://anonscm.debian.org/cgit/pkg-mutt/mutt.git/tree/debian/patches?h=experimental for the patch list (https://anonscm.debian.org/cgit/pkg-mutt/mutt.git/tree/debian/patches/series?h=experimental is the series file). So far, in terms of features, we lack:
Thanks for all of your efforts :) |
|
Great stuff! I'll look through the list of patches soon (it's very late now).
Got, got, got. I think I got my Debian patches from Sid.
And thanks for yours. |
|
I've already used these from your branch and they were very helpful :) At this point, I'm interested in either seeing these get upstreamed to neomutt (and perhaps cleaned up in the process) or to drop them from the package. Debian has been historically accepting upstream patches in the package but nowadays that we have a reasonable upstream (you ;), we should get back to what our job is really about: packaging and integration. |
|
Can we close this now? Or do we still have some pending patches from Debian that can be applied? |
|
Leave it for now, please. |
|
I've committed two new branches and merged them into neomutt. They are I've had a good look at the third patch you wanted, sensible-browser, but it needs more thought. |
|
Thanks thanks thanks for timeout!
Richard
Sent by mobile; excuse my brevity.
|
First of, let me say thank you so much for neomutt. I've been disappointed by the state of mutt for many years now and I find this project to be very refreshing to see and very much appreciated.
As you may probably be aware, Debian, like most downstreams, carries a lot of patches on top of mutt. In Debian's case, a few of them (like the sidebar and a few others) are distributed separately as a "mutt-patched" binary package, generated from the same source. The "stock" mutt package is also heavily patched; e.g. compressed folders is part of it (the line isn't very clear and I think it been based traditionally on stability merits). mutt-kz is also part of Debian, but it's generated from a separate source package and carries its own set of patches; I won't be focusing on that here.
Below you'll find the Debian patches, as can be found in pkg-mutt's git, split by their category, and with their status documented:
features
misc
upstream
mutt-patched (only included in the "mutt-patched" binary package)
Debian-specific
Legend:
I'm not part of the Debian mutt maintainer team (yet :), but I am a Debian Developer, familiar with how Debian works and was briefly in touch with the mutt maintainers earlier this week to discuss neomutt. My recommendation to them was to replace our mutt-patched package (and in the future, mutt-kz as well) straight with neomutt.
For this to happen, we'd need to forward-port all of the above patches that have no neomutt equivalent — and in the case of sensible-browser, compressed folders and NNTP, possibly replace them with your own cleaned-up versions from the devel/ branches. The only ones that don't seem to be applying cleanly so far are the multiple FCC and timeout-hook ones. Hopefully these can found their way to neomutt?
Ideally, now that there is a responsive upstream, we wouldn't have to carry any upstream patches in the Debian package. I hope the above helps you get a few more patches on your radar so that we can avoid dealing with merge-conflicts ourselves and that others can benefit from these :)
The text was updated successfully, but these errors were encountered: