| View previous topic :: View next topic |
| Should it be adopted? |
| Yes - we want speed |
|
25% |
[ 5 ] |
| No - we like slow portage |
|
75% |
[ 15 ] |
|
| Total Votes : 20 |
|
| Author |
Message |
NeddySeagoon
Administrator
Joined: 05 Jul 2003
Posts: 54234
Location: 56N 3W
|
 Posted: Wed Mar 07, 2018 9:44 am Post subject: |
 |
|
The_Document,
The poll, as worded, is a little like buying oats.
| Quote: | | “If you want nice fresh oats you have to pay a fair price. If you can be satisfied with oats that have already been through the horse, that comes a little cheaper.” |
Portage works today. Package core does not - yet.
That does not prevent Gentoo using bits of Package core in tooling, where its
a) up to the job and
b) faster than Portage.
_________________
Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
| Back to top |
|
 |
ilnanny
Tux's lil' helper
Joined: 17 Jul 2017
Posts: 123
Location: Italy/Taranto
|
|
| Back to top |
|
|
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Thu Mar 08, 2018 12:41 am Post subject: |
|
|
| ilnanny wrote: | | I'd just be happy with portage being multithreaded |
--jobs=$(nproc) not good enough? |
|
| Back to top |
|
|
The Doctor
Moderator
Joined: 27 Jul 2010
Posts: 2678
|
| Posted: Thu Mar 08, 2018 2:48 am Post subject: |
|
|
If I recall the resource hogs of portage are tasks that cannot be multithreaded. I haven't worked on the project so I'm not 100% sure on that.
_________________
First things first, but not necessarily in that order.
Apologies if I take a while to respond. I'm currently working on the dematerialization circuit for my blue box. |
|
| Back to top |
|
|
tld
Veteran
Joined: 09 Dec 2003
Posts: 1816
|
| Posted: Thu Mar 08, 2018 2:50 pm Post subject: |
|
|
| The Doctor wrote: | | If I recall the resource hogs of portage are tasks that cannot be multithreaded. I haven't worked on the project so I'm not 100% sure on that. |
I haven't either, but the biggest culprit is that world update dependency check, and I'd be very surprised if that could be.
Tom |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Sun Mar 11, 2018 3:23 am Post subject: |
|
|
I voted "yes" because I know from long experience that pkgcore is blazingly fast (especially at the large "world update" dependency check.)
It's a shame it fell behind from EAPI5 onward, and I'd love to see it back up to par.
For all those who think portage can simply steal algorithms from pkgcore, the major speedups in portage a few years ago were imports from pkgcore (frozen_sets was one.)
Any more than that requires a rewrite: which is exactly what pkgcore is, from the previous maintainer of portage.
So yes, there's still cross-collaboration, which is natural given that ferringb handed portage over to zmedico, so he could focus on "portage2" or pkgcore.
As mentioned, snakeoil is the underlying core implementation, all in C.
You simply will not get a better implementation out of portage (or much else, imo.)
WRT ease of usage, it's definitely as easy as portage; it has switches so its output is compatible, to the extent that we can script against it using the same code. Again, I have years of experience doing just that in update, so I know this to be true.
Seriously, anyone who can help out, /join #pkgcore (or #gentoo-pkgcore it used to be), start testing it in a throwaway chroot or VM if you can, and file bug reports and hopefully patches at some point. |
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3343
Location: Rasi, Finland
|
| Posted: Sun Mar 11, 2018 10:10 am Post subject: |
|
|
| steveL wrote: | As mentioned, snakeoil is the underlying core implementation, all in C.
You simply will not get a better implementation out of portage (or much else, imo.) |
Paludis. It doesn't depend on Python, which I think is great. But it's fully C++ not C.
_________________
..: Zucca :..
Gentoo IRC channels reside on Libera.Chat.
--
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Sun Mar 11, 2018 10:30 am Post subject: |
|
|
| steveL wrote: | As mentioned, snakeoil is the underlying core implementation, all in C.
You simply will not get a better implementation out of portage (or much else, imo.) |
| Zucca wrote: | | Paludis. It doesn't depend on Python, which I think is great. But it's fully C++ not C. |
That was shown to be a busted flush several years ago.
It started out with claims for speed, but those were quickly debunked, whereupon it became all about "correctness" (as decreed by McCreesh, however deluded his thinking clearly was and no doubt remains.)
Patrick (bonsaikitten) used to do speed comparisons every year or so, just for a laugh. With portage. |
|
| Back to top |
|
|
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Sun Mar 11, 2018 7:10 pm Post subject: |
|
|
| steveL wrote: | That was shown to be a busted flush several years ago.
It started out with claims for speed, but those were quickly debunked, whereupon it became all about "correctness" (as decreed by McCreesh, however deluded his thinking clearly was and no doubt remains.)
Patrick (bonsaikitten) used to do speed comparisons every year or so, just for a laugh. With portage. |
I used it for longer than a reasonable amount of time so I can vouch for that. `cave resolve -cx world` would take over half an hour sometimes, even with the optional and badly documented write cache functioning.
The only nice features it had (config dirs, repos.conf, phase hooks, automatic blocker resolution) are all in portage now. I defenestrated the entire thing eventually after reporting a fairly obvious bug (a ".." symlink in src_install caused it to fill up the disk in an endless loop) only to have it wontfix'ed with the excuse that the PM Spec - which he also wrote - doesn't disallow it. |
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3343
Location: Rasi, Finland
|
| Posted: Sun Mar 11, 2018 8:08 pm Post subject: |
|
|
| Ant P. wrote: | I used it for longer than a reasonable amount of time so I can vouch for that. `cave resolve -cx world` would take over half an hour sometimes, even with the optional and badly documented write cache functioning.
The only nice features it had (config dirs, repos.conf, phase hooks, automatic blocker resolution) are all in portage now. |
Well. I won't be testing Paludis out any time soon then.  | Ant P. wrote: | | I defenestrated the entire thing eventually after reporting a fairly obvious bug (a ".." symlink in src_install caused it to fill up the disk in an endless loop) only to have it wontfix'ed with the excuse that the PM Spec - which he also wrote - doesn't disallow it. |
"It's not a bug, it's a feature" -attitude at its worst. I think I take back my last comment... I think I won't be trying out Paludis ever, unless it's forked.
Anyways. I write my ebuilds with EAPI 6. So, Portage it is.
_________________
..: Zucca :..
Gentoo IRC channels reside on Libera.Chat.
--
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Sun Mar 11, 2018 10:34 pm Post subject: |
|
|
| The only thing I'll say in its defence is the config file layout in /etc/ was way more rational and tab-friendly. I've started putting ad-hoc symlinks in /etc/portage/ because typing p<tab><tab>a<tab>c<tab>use<enter> every time was driving me insane. |
|
| Back to top |
|
|
Mr. T.
Guru
Joined: 26 Dec 2016
Posts: 477
|
| Posted: Tue Mar 13, 2018 3:46 pm Post subject: |
|
|
I was reading a manual when I read a note about the literate programming. This manual explicitly mentions the article on Wikipedia.
I found the article valid and then I searched the application named literate in the Portage tree and I also searched related softwares.
| Code: | larry $ emerge -s literate
app-vim/alternate |
| Goverp wrote: | If pkgcore has algorithms that perform better than those in portage, while offering the same function, I'd hope portage stole them.
Even so, having an alternative development to portage is good, as it's a way to explore different algorithms and user interfaces. [...] |
| Code: | larry $ pquery --no-version --unfiltered --attr description -S literate
app-editors/leo description="Leo: Literate Editor with Outline"
app-text/noweb description="a Literate programming tool, lighter than web"
dev-haskell/markdown-unlit description="Literate Haskell support for Markdown"
dev-ml/ocamlweb description="O'Caml literate programming tool"
dev-perl/Lingua-Translit description="Transliterates text between writing system"
dev-php/pecl-translit description="Transliterates non-latin character sets to latin"
www-apps/gitit description="Wiki using happstack, git or darcs, and pandoc"
x11-misc/rodent description="A fast, small and powerful file manager and graphical shell" |
It is obvious that there is a bug somewhere. In short, I hope the algorithms are tested before being integrated elsewhere. |
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3343
Location: Rasi, Finland
|
| Posted: Tue Mar 13, 2018 8:52 pm Post subject: |
|
|
| helecho wrote: | | Code: | larry $ emerge -s literate
app-vim/alternate |
| Code: | larry $ pquery --no-version --unfiltered --attr description -S literate
app-editors/leo description="Leo: Literate Editor with Outline"
app-text/noweb description="a Literate programming tool, lighter than web"
dev-haskell/markdown-unlit description="Literate Haskell support for Markdown"
dev-ml/ocamlweb description="O'Caml literate programming tool"
dev-perl/Lingua-Translit description="Transliterates text between writing system"
dev-php/pecl-translit description="Transliterates non-latin character sets to latin"
www-apps/gitit description="Wiki using happstack, git or darcs, and pandoc"
x11-misc/rodent description="A fast, small and powerful file manager and graphical shell" |
It is obvious that there is a bug somewhere. In short, I hope the algorithms are tested before being integrated elsewhere. |
Let me try: | shellcmd: emerge -qS literate: |
[ Results for search key : literate ]
Searching...
* app-doc/docco
* app-editors/leo
* app-text/noweb [ Masked ]
* app-vim/alternate
* dev-haskell/markdown-unlit
* dev-ml/ocamlweb
* dev-perl/Lingua-Translit
* dev-php/pecl-translit
[ Applications found : 8 ] |
... But don't hold your breath. I believe eix is better at searching descs.
I wonder if the spec search could be multithreaded...
Anyway. I believe searching with emerge -s does list results that are close to the search string (in case of a typo, for example).
_________________
..: Zucca :..
Gentoo IRC channels reside on Libera.Chat.
--
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
asturm
Developer
Joined: 05 Apr 2007
Posts: 8935
|
| Posted: Tue Mar 13, 2018 8:57 pm Post subject: |
|
|
| Are you really just confusing emerge -s with emerge -S? |
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3343
Location: Rasi, Finland
|
| Posted: Tue Mar 13, 2018 10:11 pm Post subject: |
|
|
| asturm wrote: | | Are you really just confusing emerge -s with emerge -S? |
Probably is.
I still have to wonder where does pquery gets those few extra hits...
_________________
..: Zucca :..
Gentoo IRC channels reside on Libera.Chat.
--
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
The Main Man
Veteran
Joined: 27 Nov 2014
Posts: 1166
Location: /run/user/1000
|
| Posted: Tue Mar 13, 2018 11:54 pm Post subject: |
|
|
I guess it depends on overlays, local repos etc.
In my case there are only 4 results , it must be that then 
| Code: | $ emerge -qS literate
[ Results for search key : literate ]
Searching...
* app-editors/leo
* app-vim/alternate
* dev-haskell/markdown-unlit
* dev-perl/Lingua-Translit
[ Applications found : 4 ] |
eix is better though
| Code: | $ eix -S literate
* app-editors/leo
Available versions: 5.6 {doc PYTHON_TARGETS="python2_7"}
Homepage: https://github.com/leo-editor/leo-editor/
Description: Leo: Literate Editor with Outlines
* app-text/noweb
Available versions: [M]2.11b-r4 {emacs examples}
Homepage: https://www.cs.tufts.edu/~nr/noweb/
Description: a literate programming tool, lighter than web
* dev-haskell/markdown-unlit
Available versions: 0.2.0.1(0/0.2.0.1) {doc hscolour profile test}
Homepage: http://hackage.haskell.org/package/markdown-unlit
Description: Literate Haskell support for Markdown
* dev-ml/ocamlweb
Available versions: (~)1.39
Homepage: http://www.lri.fr/~filliatr/ocamlweb/
Description: O'Caml literate programming tool
* dev-perl/Lingua-Translit
Available versions: 0.250.0 (~)0.270.0 (~)0.280.0
Homepage: http://search.cpan.org/dist/Lingua-Translit/
Description: Transliterates text between writing systems
* dev-php/pecl-translit
Available versions: (~)0.6.2-r1 {PHP_TARGETS="php5-6 php7-0"}
Homepage: https://github.com/derickr/pecl-translit
Description: Transliterates non-latin character sets to latin
Found 6 matches
|
|
|
| Back to top |
|
|
radhermit
Developer
Joined: 15 Apr 2010
Posts: 9
|
| Posted: Wed Mar 14, 2018 12:44 am Post subject: |
|
|
| helecho wrote: | | It is obvious that there is a bug somewhere. In short, I hope the algorithms are tested before being integrated elsewhere. |
As with most pkgcore-based tools, don't assume it's 100% compatible with its portage analog. In this case the full description for `pquery -S` is as follows: regexp search on description and longdescription.
In this case the extra matches come from longdesc matches in the metadata.xml. |
|
| Back to top |
|
|
Mr. T.
Guru
Joined: 26 Dec 2016
Posts: 477
|
| Posted: Wed Mar 14, 2018 8:46 am Post subject: |
|
|
| I read the description (for option S) but I didn't know the meaning of "longdescription". This precision makes a significant difference, thank! |
|
| Back to top |
|
|
|
Gentoo Chat
