Binariz, I will send you a working release which I use an windows and svn when I am back at home thus weekend an I try to commit a working script a bit later after some improvements.
Greetings xqt
----- Ursprüngliche Nachricht ----- Von: Bináris Gesendet: 11.01.2014 11:40 An: Pywikipedia discussion list Betreff: [Pywikipedia-l] Unknown command git
Hi!
Az last -- as SVN is totally broken -- I downloaded the latest nightly of compat. First problem: whatever script I try to run DOS (Windows command) writes me the thing in subject. I think I have nothing to do with this stuff called git at the moment and didn't say to Windows to run anything by this name. Why I am damned with these messages and how can I prevent them? At the moment I just want to use my bot.
Second problem: version.py from latest nightly says: Pywikibot: wikipedia.py (r-1 (unknown), b1e1a57, 2014/01/01, 13:01:20, OUTDATED) Release version: 1.0b1
Why?
Yes, I really AM very angry and frustrated with this whole mess and I wish we have never thrown away a well working system just for nothing. Once I used to be a developer of this framework, now I have troubles with making it run at all. I feel I have to build and learn everything again from scratch what I had built and learnt for years. Arghhhxcnmv^^ˇ#&>@&{˘°~˘^°~˘^
Hi.
Quick hach waiting for Xqt.
In pywikibot/version.py: try: #(tag, rev, date, hsh) = getversion_git() raise Exception
Bye Mpaa
On Sat, Jan 11, 2014 at 12:15 PM, info@gno.de wrote:
Binariz, I will send you a working release which I use an windows and svn when I am back at home thus weekend an I try to commit a working script a bit later after some improvements.
Greetings xqt
Von: Bináris Gesendet: 11.01.2014 11:40 An: Pywikipedia discussion list Betreff: [Pywikipedia-l] Unknown command git
Hi!
Az last -- as SVN is totally broken -- I downloaded the latest nightly of compat. First problem: whatever script I try to run DOS (Windows command) writes me the thing in subject. I think I have nothing to do with this stuff called git at the moment and didn't say to Windows to run anything by this name. Why I am damned with these messages and how can I prevent them? At the moment I just want to use my bot.
Second problem: version.py from latest nightly says: Pywikibot: wikipedia.py (r-1 (unknown), b1e1a57, 2014/01/01, 13:01:20, OUTDATED) Release version: 1.0b1
Why?
Yes, I really AM very angry and frustrated with this whole mess and I wish we have never thrown away a well working system just for nothing. Once I used to be a developer of this framework, now I have troubles with making it run at all. I feel I have to build and learn everything again from scratch what I had built and learnt for years. Arghhhxcnmv^^ˇ#&>@&{˘°~˘^°~˘^
-- Bináris
Pywikipedia-l mailing list Pywikipedia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/pywikipedia-l
2014/1/11 Mpaa mpaa.wiki@gmail.com
Hi.
Quick hach waiting for Xqt.
In pywikibot/version.py: try: #(tag, rev, date, hsh) = getversion_git() raise Exception
Thank you! That works, but scripts still start slowly. I am not sure if there is a connection. Will that cause problems by update?
2014/1/11 info@gno.de
Binariz, I will send you a working release which I use an windows and svn when I am back at home thus weekend an I try to commit a working script a bit later after some improvements.
Thank you in advance! P.s. 's' in Bináris sounds like 'sh' in English or short 'sch' in German. :-) It means *binary.*
I will have to agree with Bináris, Ive hacked my local checkout to avoid the recent fuckups that have been added. I basically killed the whole externals import process and several other things. , On Sat, Jan 11, 2014 at 6:44 AM, Bináris wikiposta@gmail.com wrote:
2014/1/11 info@gno.de
Binariz, I will send you a working release which I use an windows and svn
when I am back at home thus weekend an I try to commit a working script a bit later after some improvements.
Thank you in advance! P.s. 's' in Bináris sounds like 'sh' in English or short 'sch' in German. :-) It means *binary.*
Pywikipedia-l mailing list Pywikipedia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/pywikipedia-l
2014/1/11 John phoenixoverride@gmail.com
I will have to agree with Bináris, Ive hacked my local checkout to avoid the recent fuckups that have been added. I basically killed the whole externals import process and several other things.
Could you please create a howto? It could be placed under PWB documentation as an unofficial extension with the template "Don't try this at home except you want to try it". :-)
On Sat, Jan 11, 2014 at 6:52 AM, John phoenixoverride@gmail.com wrote:
I will have to agree with Bináris, Ive hacked my local checkout to avoid the recent fuckups that have been added. I basically killed the whole externals import process and several other things.
Presumably you filed bug reports as well, right? Otherwise it's pretty unlikely things will improve.
Tom
I am glad I'n not the only one who have problem with damned git. Now I use SVM again, and the only problem is when I update after many versions, that there are conflicts between local and global files.
JAnD
---
2014/1/11 Tom Morris tfmorris@gmail.com
On Sat, Jan 11, 2014 at 6:52 AM, John phoenixoverride@gmail.com wrote:
I will have to agree with Bináris, Ive hacked my local checkout to avoid the recent fuckups that have been added. I basically killed the whole externals import process and several other things.
Presumably you filed bug reports as well, right? Otherwise it's pretty unlikely things will improve.
Tom
Pywikipedia-l mailing list Pywikipedia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/pywikipedia-l
Given the support and feedback that I have received going back to the horribly broken addition by Trigon of a logging system Filling bugs is useless. Here is a quote from IRC earlier today:
<Me> having the framework check git on every script invoke is assine <valhallasw> No, it's not. <valhallasw> There's basically no other way to get the current checkout version. <Me> I guess you havent heard of caching that information and only checking ever X days? <valhallasw> why would that be a sensible strategy? <valhallasw> because that would add a lot of code for exactly zero gain
Per valhallasw's request I did some profiling, They quoted a 0.37 seconds run-time to invoke "interwiki.py -help" I ran the same tests, both with a clean git checkout and my modified checkout. 0.389 seconds (mine) vs 1.257 that is a factor of 3 in regards to execution time (all I really did was remove the version checks on every script invoke) Besides the execution time it adds an additional ~500 hits per day to the git servers for my scheduled tasks. Extrapolate that across the number of operators that we have, and thats no small chunk of change. I also got a nice lecture about toning my temper down, for calling out these rather significant bad ideas. I have kept a very cool head and my temper is a non-issue, I just call a spade a spade.
Besides those issues, the decision to make the incomplete and broken "rewrite" into "core" and shelve the easier to use and more stable trunk into "compat" was an issue that wont get resolved. Ive tried several times to get it setup on windows and guess what? it works about as well as a square tire. I spent about 30 seconds and found 6 issues that make the branch unusable on windows. I have lost count of the number of people who have come into irc://freenode/pywikipediabot asking for help setting up core and have been unable. Setting up compat? takes less than 5 instructions and I have yet to have someone unable to do it. On core? Ive had maybe 3 be able to do it.
Switching to git wasnt that big of an issue, however the forced use of gerrit is. Gerrit has no usable windows based interface. and then your code gets -2'ed because the reviewer has no clue about the changes being made and instead of asking just gives a comment that shows they are reviewing code that they know zero about. Having to use the gerrit patch uploader to even get things submitted is ugly as I got -2'ed for PEP8 spacing issues that where correct in my code, but between git and the patch uploader somehow got eaten.
On Sat, Jan 11, 2014 at 4:57 PM, Tom Morris tfmorris@gmail.com wrote:
On Sat, Jan 11, 2014 at 6:52 AM, John phoenixoverride@gmail.com wrote:
I will have to agree with Bináris, Ive hacked my local checkout to avoid the recent fuckups that have been added. I basically killed the whole externals import process and several other things.
Presumably you filed bug reports as well, right? Otherwise it's pretty unlikely things will improve.
Tom
Pywikipedia-l mailing list Pywikipedia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/pywikipedia-l
On 11 January 2014 23:28, John phoenixoverride@gmail.com wrote:
Given the support and feedback that I have received going back to the horribly broken addition by Trigon of a logging system Filling bugs is useless. Here is a quote from IRC earlier today:
(...)
Per valhallasw's request I did some profiling, They quoted a 0.37 seconds run-time to invoke "interwiki.py -help" I ran the same tests, both with a clean git checkout and my modified checkout. 0.389 seconds (mine) vs 1.257 that is a factor of 3 in regards to execution time (all I really did was remove the version checks on every script invoke)
Besides the execution time it adds an additional ~500 hits per day to the
git servers for my scheduled tasks.
I still really don't get the issue with a second. Sure, it could be faster. Sure, it could be more efficient. Sure, we could save git.wikimedia.orgsome requests (but I have not had complaints from the WMF, so it's probably not an issue). But I really cannot be bothered spending my time to fix this for you. As I mentioned, you are more than welcome to submit a patch, but maybe not the one which you yourself described as "[a] fix [that] has about as much chance of being accepted as a snowball in hell". I'm not opposed to caching, nor to adding a user-config option to remove version checking.
I also got a nice lecture about toning my temper down, for calling out
these rather significant bad ideas. I have kept a very cool head and my temper is a non-issue, I just call a spade a spade.
I am, however, opposed to calling changes 'obnoxious fuckup's.
Besides those issues, the decision to make the incomplete and broken
"rewrite" into "core" and shelve the easier to use and more stable trunk into "compat" was an issue that wont get resolved.
Ive tried several times to get it setup on windows and guess what? it works
about as well as a square tire. I spent about 30 seconds and found 6 issues that make the branch unusable on windows.
As Tom mentioned: have you created bugs for these? Because I have absolutely zero issues with core on windows, and I've been using it without any issues on linux for ages.
I have lost count of the number of people who have come into irc://freenode/pywikipediabot asking for help setting up core and have been unable. Setting up compat? takes less than 5 instructions and I have yet to have someone unable to do it. On core? Ive had maybe 3 be able to do it.
Well, then core clearly is better, with the one-liner 'pwb.py login'. But seriously, I have no clue what problems you encounter. Download python, download a nightly, run a bot. Nothing has changed in that respect.
and then your code gets -2'ed because the reviewer has no clue about the changes being made and instead of asking just gives a comment that shows they are reviewing code that they know zero about.
I sincerely doubt that.
Having to use the gerrit patch uploader to even get things submitted is ugly as I got -2'ed for PEP8 spacing issues that where correct in my code, but between git and the patch uploader somehow got eaten.
And again, please open a bug report, as this should not happen: https://github.com/valhallasw/gerrit-patch-uploader/issues
Merlijn
Seriously? you need to can the bullshit. I posted a patch and I got a response and code review of:
I can't see why you're doing this, when can set protocol in so many other ways (which we did)
Which is utter bullshit. Until I provided the patch there was a hole in the LinkSearch code, and it was broken Ill re-download core and prove that its a piece of shit still.
Your response of Oh yes I know the code is horribly written and broken but I want that feature and I cant be bothered to fix it, is the reason I dont waste my time with a patch. This reminds me of what happened on en,wikipedia and visual editor, the WMF forced a broken product out regardless of the issues, It took a revolt and admins forcibly disabling the broken extension via common.js before they started to listen. I guess thats where this is headed.
On Sat, Jan 11, 2014 at 7:11 PM, Merlijn van Deen valhallasw@arctus.nlwrote:
On 11 January 2014 23:28, John phoenixoverride@gmail.com wrote:
Given the support and feedback that I have received going back to the horribly broken addition by Trigon of a logging system Filling bugs is useless. Here is a quote from IRC earlier today:
(...)
Per valhallasw's request I did some profiling, They quoted a 0.37 seconds run-time to invoke "interwiki.py -help" I ran the same tests, both with a clean git checkout and my modified checkout. 0.389 seconds (mine) vs 1.257 that is a factor of 3 in regards to execution time (all I really did was remove the version checks on every script invoke)
Besides the execution time it adds an additional ~500 hits per day to the
git servers for my scheduled tasks.
I still really don't get the issue with a second. Sure, it could be faster. Sure, it could be more efficient. Sure, we could save git.wikimedia.org some requests (but I have not had complaints from the WMF, so it's probably not an issue). But I really cannot be bothered spending my time to fix this for you. As I mentioned, you are more than welcome to submit a patch, but maybe not the one which you yourself described as "[a] fix [that] has about as much chance of being accepted as a snowball in hell". I'm not opposed to caching, nor to adding a user-config option to remove version checking.
I also got a nice lecture about toning my temper down, for calling out
these rather significant bad ideas. I have kept a very cool head and my temper is a non-issue, I just call a spade a spade.
I am, however, opposed to calling changes 'obnoxious fuckup's.
Besides those issues, the decision to make the incomplete and broken
"rewrite" into "core" and shelve the easier to use and more stable trunk into "compat" was an issue that wont get resolved.
Ive tried several times to get it setup on windows and guess what? it
works about as well as a square tire. I spent about 30 seconds and found 6 issues that make the branch unusable on windows.
As Tom mentioned: have you created bugs for these? Because I have absolutely zero issues with core on windows, and I've been using it without any issues on linux for ages.
I have lost count of the number of people who have come into irc://freenode/pywikipediabot asking for help setting up core and have been unable. Setting up compat? takes less than 5 instructions and I have yet to have someone unable to do it. On core? Ive had maybe 3 be able to do it.
Well, then core clearly is better, with the one-liner 'pwb.py login'. But seriously, I have no clue what problems you encounter. Download python, download a nightly, run a bot. Nothing has changed in that respect.
and then your code gets -2'ed because the reviewer has no clue about the changes being made and instead of asking just gives a comment that shows they are reviewing code that they know zero about.
I sincerely doubt that.
Having to use the gerrit patch uploader to even get things submitted is ugly as I got -2'ed for PEP8 spacing issues that where correct in my code, but between git and the patch uploader somehow got eaten.
And again, please open a bug report, as this should not happen: https://github.com/valhallasw/gerrit-patch-uploader/issues
Merlijn
Pywikipedia-l mailing list Pywikipedia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/pywikipedia-l
Fucked up, just like I thought, I did a git clone, ran pwb,py login and guess what? shocker traceback
Please see http://www.mediawiki.org/wiki/Manual:Pywikipediabot/Windows Traceback (most recent call last): File "C:\rewrite\pwb.py", line 123, in <module> tryimport_pwb() File "C:\rewrite\pwb.py", line 30, in tryimport_pwb import pywikibot File "C:\rewrite\pywikibot__init__.py", line 412, in <module> from .page import Page, ImagePage, Category, Link, User, ItemPage, PropertyP age, Claim File "C:\rewrite\pywikibot\page.py", line 17, in <module> import pywikibot.site File "C:\rewrite\pywikibot\site.py", line 32, in <module> from pywikibot import pagegenerators File "C:\rewrite\pywikibot\pagegenerators.py", line 31, in <module> from pywikibot.comms import http File "C:\rewrite\pywikibot\comms\http.py", line 35, in <module> import queue as Queue ImportError: No module named queue
oython 2.7.3 pwb.py login
On Sun, Jan 12, 2014 at 3:32 AM, Wieland Hoffmann themineo@gmail.comwrote:
Hallo John:
Fucked up, just like I thought, I did a git clone, ran pwb,py login and guess what?
Can you please copy&paste the exact command you ran and which Python version you're using?
-- Wieland
Pywikipedia-l mailing list Pywikipedia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/pywikipedia-l
On Sun, Jan 12, 2014 at 6:43 AM, Wieland Hoffmann themineo@gmail.comwrote:
Hallo John:
oython 2.7.3 pwb.py login
Sorry, I meant the command you used to clone the repository.
-- Wieland
Pywikipedia-l mailing list Pywikipedia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/pywikipedia-l
[the mail I'm replying to contained only an image]
Hallo John:
I'm not sure why you expect that checkout to work: [0] clearly says
git clone --recursive https://gerrit.wikimedia.org/r/pywikibot/core.git
(note the "--recursive" part). Your checkout will obviously not work if you don't check the "recursive" checkbox. The alternative to getting a working checkout is doing a non-recursive clone and then doing
git submodule update --init
as described in the README file at [1].
[0] https://www.mediawiki.org/wiki/Manual:Pywikipediabot/Installation#Download_P... [1] https://git.wikimedia.org/blob/pywikibot%2Fcore.git/cd5716cdd06a8347212c63cf...
On 12 January 2014 13:12, Wieland Hoffmann themineo@gmail.com wrote:
I'm not sure why you expect that checkout to work: [0] clearly says
git clone --recursive https://gerrit.wikimedia.org/r/pywikibot/core.git
(note the "--recursive" part). Your checkout will obviously not work if you don't check the "recursive" checkbox. The alternative to getting a working checkout is doing a non-recursive clone and then doing
Note that compat also doesn't work without --recursive, as i18n would be missing. However, I think we *should* show a clearer error message. See https://bugzilla.wikimedia.org/show_bug.cgi?id=59970
Merlijn
Hi John,
Sorry, I do not know how I can help you with this statement. Neither on coding stuff nor on our workflow. Ok your are frustrated and angry on some issues. But what can we improve in your opinion?
Greetings xqt
----- Original Nachricht ---- Von: John phoenixoverride@gmail.com An: Pywikipedia discussion list pywikipedia-l@lists.wikimedia.org Datum: 12.01.2014 01:30 Betreff: Re: [Pywikipedia-l] Unknown command git
Seriously? you need to can the bullshit. I posted a patch and I got a response and code review of:
I can't see why you're doing this, when can set protocol in so many other ways (which we did)
Which is utter bullshit. Until I provided the patch there was a hole in the LinkSearch code, and it was broken Ill re-download core and prove that its a piece of shit still.
Your response of Oh yes I know the code is horribly written and broken but I want that feature and I cant be bothered to fix it, is the reason I dont waste my time with a patch. This reminds me of what happened on en,wikipedia and visual editor, the WMF forced a broken product out regardless of the issues, It took a revolt and admins forcibly disabling the broken extension via common.js before they started to listen. I guess thats where this is headed.
On Sat, Jan 11, 2014 at 7:11 PM, Merlijn van Deen valhallasw@arctus.nlwrote:
On 11 January 2014 23:28, John phoenixoverride@gmail.com wrote:
Given the support and feedback that I have received going back to the horribly broken addition by Trigon of a logging system Filling bugs is useless. Here is a quote from IRC earlier today:
(...)
Per valhallasw's request I did some profiling, They quoted a 0.37 seconds run-time to invoke "interwiki.py -help" I ran the same tests,
both
with a clean git checkout and my modified checkout. 0.389 seconds (mine)
vs
1.257 that is a factor of 3 in regards to execution time (all I really
did
was remove the version checks on every script invoke)
Besides the execution time it adds an additional ~500 hits per day to the
git servers for my scheduled tasks.
I still really don't get the issue with a second. Sure, it could be faster. Sure, it could be more efficient. Sure, we could save git.wikimedia.org some requests (but I have not had complaints from the WMF, so it's probably not an issue). But I really cannot be bothered spending my time to fix this for you. As
I
mentioned, you are more than welcome to submit a patch, but maybe not the one which you yourself described as "[a] fix [that] has about as much chance of being accepted as a snowball in hell". I'm not opposed to caching, nor to adding a user-config option to remove version checking.
I also got a nice lecture about toning my temper down, for calling out
these rather significant bad ideas. I have kept a very cool head and my temper is a non-issue, I just call a spade a spade.
I am, however, opposed to calling changes 'obnoxious fuckup's.
Besides those issues, the decision to make the incomplete and broken
"rewrite" into "core" and shelve the easier to use and more stable trunk into "compat" was an issue that wont get resolved.
Ive tried several times to get it setup on windows and guess what? it
works about as well as a square tire. I spent about 30 seconds and found
6
issues that make the branch unusable on windows.
As Tom mentioned: have you created bugs for these? Because I have absolutely zero issues with core on windows, and I've been using it
without
any issues on linux for ages.
I have lost count of the number of people who have come into irc://freenode/pywikipediabot asking for help setting up core and have been unable. Setting up compat? takes less than 5 instructions and I
have
yet to have someone unable to do it. On core? Ive had maybe 3 be able to
do
it.
Well, then core clearly is better, with the one-liner 'pwb.py login'. But seriously, I have no clue what problems you encounter. Download python, download a nightly, run a bot. Nothing has changed in that respect.
and then your code gets -2'ed because the reviewer has no clue about the changes being made and instead of asking just gives a comment that shows they are reviewing code that they know zero about.
I sincerely doubt that.
Having to use the gerrit patch uploader to even get things submitted is ugly as I got -2'ed for PEP8 spacing issues that where correct in my
code,
but between git and the patch uploader somehow got eaten.
And again, please open a bug report, as this should not happen: https://github.com/valhallasw/gerrit-patch-uploader/issues
Merlijn
Pywikipedia-l mailing list Pywikipedia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/pywikipedia-l
Pywikipedia-l mailing list Pywikipedia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/pywikipedia-l
pywikipedia-l@lists.wikimedia.org