| Summary: | sys-apps/paludis: last rites | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Michał Górny <mgorny> |
| Component: | Current packages | Assignee: | Michał Górny <mgorny> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | adia, bfg, bitozoid, denisgolovan, eugene.shalygin, hendrik, hypnos75, I.zaufi, xdudka00 |
| Priority: | Normal | Keywords: | PMASKED |
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| URL: | https://archives.gentoo.org/gentoo-dev/message/e14746e37198ed8310f30fe2be168fb0 | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Bug Depends on: | 655976, 658268, 658270, 658272, 658274, 658276 | ||
| Bug Blocks: | |||
|
Description
Michał Górny
2018-06-17 09:03:36 UTC
For the record, I'm open to the idea of having an official Gentoo fork of Paludis. However, this would require some extra manpower actually working on the patches. I have been using paludis to keep my Gentoo systems up2date since 2008 without any major issues. I do not think that any of the linked bugs is serious enough to warrant immediate removal of the paludis package from Gentoo. I am not saying that I have enough time and skills to maintain it myself but I would be happy to continue using it as it is until a more serious bug appears. (In reply to Kamil Dudka from comment #2) > I have been using paludis to keep my Gentoo systems up2date since 2008 Same here. >10 years of using `paludis` + a bunch of my own hooks give me full control over my system. I don't want to lose it and use `emerge`... again... I'm going to try Exherbo this weekend w/ few my Gentoo friends (it'll be "Exherbo install fest" :) While Gentoo guys tell about the low quality of Exherbo packages, I see the same in Gentoo last years... so as for me, both systems are equal (but the other one won't kick out my favorite package manager from the tree) It's quite a shame that paludis is stale and not evolving, really. Using it for years now and having a blast. Only real major issue now is lack of support of EAPI=7. If this one can be fixed with some patches then paludis will live again. exherbo looks usable now, still missing few apps i need, but not too much of it. maybe it's time to move .... anyway if main paludis developers as well as gentoo team does not interested in supporting gentoo in paludis, it have no future... (In reply to Kamil Dudka from comment #2) > I have been using paludis to keep my Gentoo systems up2date since 2008 > without any major issues. I do not think that any of the linked bugs is > serious enough to warrant immediate removal of the paludis package from > Gentoo. Are you saying that inability to support new ebuilds is not 'serious enough'? Will it become serious enough if you won't be able to upgrade your system because of unsupported dependencies? Will it be serious enough when you'll be stuck with vulnerable packages because paludis does not support the upgrade to fixed versions? I was able to upgrade my Gentoo box yesterday using paludis just fine. Not sure if some official stats are available but my quick grep over the official Gentoo git repo gives me the following results: EAPI=2 ... 167 ebuilds EAPI=4 ... 1375 ebuilds EAPI=5 ... 13481 ebuilds EAPI=6 ... 19332 ebuilds EAPI=7 ... 178 ebuilds So it does not seem to be an urgent issue right now. I still hope that someone competent and motivated enough will step in before it becomes critical. In any case, removing paludis from Gentoo proactively is not going to help anybody IMO. It is also not clear what the non-technical issues behind the lack of interest about Gentoo in paludis community really are. Maybe it could help if they were involved in the review process when new EAPI is being approved? Just guessing. (In reply to Kamil Dudka from comment #7) > I was able to upgrade my Gentoo box yesterday using paludis just fine. Not > sure if some official stats are available but my quick grep over the > official Gentoo git repo gives me the following results: > > EAPI=2 ... 167 ebuilds > EAPI=4 ... 1375 ebuilds > EAPI=5 ... 13481 ebuilds > EAPI=6 ... 19332 ebuilds > EAPI=7 ... 178 ebuilds > > So it does not seem to be an urgent issue right now. Statistics are really meaningless here. It's enough that a single common enough package switches to EAPI 7 to make upgrade impossible. I don't know exact paludis behavior (it never worked for me!) but will it warn you that you're using obsolete version of a package whose corresponding ebuild has been removed? Because if you do that, then you can even end up using vulnerable versions because security won't be issuing GLSAs for ebuilds long gone. > I still hope that > someone competent and motivated enough will step in before it becomes > critical. In any case, removing paludis from Gentoo proactively is not > going to help anybody IMO. I seriously don't expect it to be usable 90 days from now. So it's better to warn people up front than wait till they learn the hard way. > > It is also not clear what the non-technical issues behind the lack of > interest about Gentoo in paludis community really are. Maybe it could help > if they were involved in the review process when new EAPI is being approved? > Just guessing. The review process is open. They used to be involved in the previous years. This year, they chose not to participate, except adding snarky comments like 'it is horrible' with no specifics. (In reply to Michał Górny from comment #8) > Statistics are really meaningless here. It's enough that a single common > enough package switches to EAPI 7 to make upgrade impossible. If it is a single package, I can take care of the upgrade myself temporarily. > I don't know exact paludis behavior (it never worked for me!) So you do not care to install paludis and paludis developers do not care to install Gentoo. It does not sound like the best way to move forward. Given the fact that many Gentoo users (including lame users like me) were able to install paludis and use it for years, it is likely not so difficult to make it work ;-) > but will it > warn you that you're using obsolete version of a package whose corresponding > ebuild has been removed? I remember it complained about packages in /var/db/pkg/world in case they could not be upgraded until I removed the entries from the file manually, or removed those packages. > I seriously don't expect it to be usable 90 days from now. So it's better > to warn people up front than wait till they learn the hard way. Warning people was definitely good course of action because otherwise I would not even know that there is something wrong with paludis in Gentoo. On the other hand, removing packages from Gentoo is not the only way to warn people. (In reply to Kamil Dudka from comment #9) > > I don't know exact paludis behavior (it never worked for me!) > > So you do not care to install paludis and paludis developers do not care to > install Gentoo. It does not sound like the best way to move forward. Given > the fact that many Gentoo users (including lame users like me) were able to > install paludis and use it for years, it is likely not so difficult to make > it work ;-) I would really appreciate if you could be kind enough to refrain from accusations like that. I did install Paludis, and I did report bugs against it. However, I have better things to do than to keep swapping system configs to workaround limitations of Paludis. Also been using paludis for 10+ years, my laptop's world has ~ 700 packages, ~ 3000 packages are installed. Everything has been working so fine and I was so busy that I hadn't caught up the news of EAPI 7, paludis devs not caring about it, etc. and I just got the news a few days ago with the mask (thanks mgorny!). First I wanted to see how portage had caught up with paludis in terms of resolver (handling deps and blockers, speed) and see if I could live without paludis -> not really, I don't know if I've become too picky but portage is unbearable with my config. I've also set up an Exherbo chroot to evaluate it, and realized after a few hours of use: - Exherbo has a lot less packages available and needs more packagers: ~ 400 of my Gentoo world packages aren't in Exherbo; some common packages (eg. texlive) are 2 years behind Gentoo. - As I'm doing some embedded software code, the cross-compilation toolchain availability and ease of install is important for me... so far I'm working around Exherbo not having anything like toolchain.eclass and cross-* categories (the stuff that make crossdev work), rather having one package built with several targets, among a fixed set, which is not cool at all -> writing my own cross-gcc recipes as a workaround. - I sometimes have to run 32-bit wine programs and I haven't checked yet whether that works fine. Anyway, since I don't think I'll be operational (finished writing my missing recipes) with Exherbo soon, I have started to work on Paludis (since yesterday) to support EAPI 7 (since my free time is limited, my goal is to have an MVP, I probably won't add testing code...) so I can keep using it on Gentoo. I'm not very familiar with the codebase (haven't done anything but small personal changes on it) but the EAPI 7 changes (except BDEPEND when doing cross-compilation of packages) don't seem too scary to handle. I (Zougloub) will be ghosting #paludis as I've done forever, so if someone wants to work on that too, we can coordinate. From the above comments (there's at least of a handful of 10+ years users), it looks like it's manageable to have a critical mass for a community to keep paludis alive on Gentoo. Having an official Gentoo fork (and proxy maintainer) may keep the spot brighter on the radar (being the loop WRT EAPI updates and not being laughed at when complaining about some portage-specific quirks that would make paludis choke) but it may not be absolutely necessary. It goes without saying that as a user, I have been aware (and I so don't care care) of all the history, the feuds, frustration and passive-aggressiveness “on both sides, on both sides”, and I just happen to appreciate paludis for its quality and the services it has been rendering me for all these years. Thankfully as of today Gentoo is OK with alternate package managers, provided their name starts with p, and we can get this to work. > However, this would require some extra manpower actually working on the patches.
Can you give some information (or a link) for minimally acceptable level of support for paludis being live on Gentoo?
I am pretty sure it's not a particularly heavy effort in fact as half of those bugs already have half-baked patches.
(In reply to MageSlayer from comment #12) > > However, this would require some extra manpower actually working on the patches. > > Can you give some information (or a link) for minimally acceptable level of > support for paludis being live on Gentoo? > I am pretty sure it's not a particularly heavy effort in fact as half of > those bugs already have half-baked patches. The bugs linked to this one. I'll try to integrate existing bug fixes starting from easy ones into https://github.com/MageSlayer/paludis-gentoo-patches I've created mirror issues for corresponding bugs. Any help will be appreciated. EAPI 7 support was not worked on, and it's actually the only requirement for paludis to be usable in Gentoo for now. (In reply to Sergey 'L29Ah' Alirzaev from comment #15) > EAPI 7 support was not worked on, and it's actually the only requirement for > paludis to be usable in Gentoo for now. I'll do my best, my free resources are limited though. For those interested - I've created an overlay for Paludis patches for Gentoo - https://github.com/MageSlayer/paludis-gentoo-overlay Thank you MageSlayer and all other contributors for setting that up. I could install your patched paludis version without any problems. Unfortunately I already encountered EAPI=7 in my attempt to update world: dev-python/flask-babel. So this confirms what Michał said and I am stuck for now. For me, paludis has always been a double-edged sword. On the downside it is unrelentingly strict about dependencies, but on the upside it also gives you an enormous amount of insight and control over the resolution process. I haven’t yet seen something like `cave resolve --explain` or `cave show -c` for portage. So when there are big conflicts with portage, I always feel kinda left alone... Of course switching back to portage would also take a lot of effort. Thanks. EAPI7 is on my TODO list. My time is limited though. (In reply to Bodo Graumann from comment #18) > I haven’t yet seen something like `cave resolve --explain` or `cave show -c` > for portage. my top rated features: * paludis hooks -- allow me to do things that impossible w/ portage (after I moved to paludis, some hacks and workarounds appeared in portage, but look ugly %) * resume file(s) -- I recall how I miss `-a` (ask) feature of `emerge`, but found resume files I won't change it back ;) * the way of multiple repositories (naturally) support... * ... and multiple profiles > Of course switching back to portage would also take a lot of effort. I've chosen another option: switch to Exherbo ;-) ... yeah, I know that it has a lack of some packages... but: * I don't know how much packages I would miss (before moving I just checked that major of them are here and fresh enough), but I'm sure I'll write `exheres` for them (all) * `paludis` is much-much faster in Exherbo -- currently I'm still preparing my system and have only 645 packages installed (it was ~1300 in my former Gentoo system), but it's incomparable faster when in Gentoo... * I wrote a dozen of `ebuild`s in my Gentoo life, but I became fun of `exheres` (and `exlibs`) after the first reading of the manual! ;-) ... already done a couple of mines ;) (already started my repo %) * if you like `-X` option of `cave` you'll wonder how different just "normal" output of Paludis in Exherbo ;-) -- thanx to some syntax capabilities (sugar?) of `exheres`... I would say it's a way better explanatory. Living in Gentoo for me sometimes was not trivial but quite fun anyway. Being a programmer I could solve any compilation problems quite easy -- it is why I love source based distros -- they give me full control over my system. As for me, last years quality of packages get lower (whatever Gentoo-fun-boys would say @ IRC/mail), so I spent some time resolving update problems, sending bug reports (which a long time could stay w/ no attention), making a patch, hacking ebuilds, maintaining my repo...I'm pretty Ok w/ that and have some fun doing it. However, all these things I can do in Exherbo as well, but with more fun and all the time sure that nobody is going to remove Paludis from my system ;-) The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=5d144f9e4df60f4e9eadb3b412e2770fb71250c3 commit 5d144f9e4df60f4e9eadb3b412e2770fb71250c3 Author: Mikle Kolyada <zlogene@gentoo.org> AuthorDate: 2018-09-19 11:27:28 +0000 Commit: Mikle Kolyada <zlogene@gentoo.org> CommitDate: 2018-09-19 11:30:16 +0000 sys-apps/paludis: remove last-rited package Closes: https://bugs.gentoo.org/658278 sys-apps/paludis/Manifest | 1 - sys-apps/paludis/metadata.xml | 17 --- sys-apps/paludis/paludis-3.0.0_pre20180214.ebuild | 148 ---------------------- sys-apps/paludis/paludis-9999.ebuild | 145 --------------------- 4 files changed, 311 deletions(-) I was able to successfully update my Gentoo boxes again with Paludis using the following overlay: https://github.com/MageSlayer/paludis-gentoo-patches/issues/6#issuecomment-423845603 |