-
Notifications
You must be signed in to change notification settings - Fork 46
My package doesn't appear in the search results anymore #397
Comments
Original comment by Anton Dries (Bitbucket: antondries, GitHub: Unknown): I have the same problem with one of my packages (problog). |
Original comment by Paddy Ganut (Bitbucket: paddy_ganut, GitHub: Unknown): This is scrambling my search results (pi3d) and seems like a very serious issue. Do any of the site maintainers have any ideas what might be causing this? PS If it helps any relevant dev. My package seemed to be correctly found as at v2.7 on 2015/08/14 but was acting up after v2.8 on 2015/11/07 |
Original comment by Senthil Muniswamy (Bitbucket: smuniswamy, GitHub: smuniswamy): Same issue with ncryptify package. Current latest version 0.0.8 is not listed upon search. Only lists the earlier versions that are supposed to be hidden. Direct link to the package shows the latest version and does not list the earlier versions. So, the issue is with the search only. |
Original comment by Brian O'Neill (Bitbucket: Twangist, GitHub: Twangist): Same issue with my package. When I search for |
Original comment by Fred Stluka (Bitbucket: fredstluka, GitHub: fredstluka): Yeah, same problem here. But I get a much more accurate answer from the script I found here: and from the command:
BTW, there are lots of people complaining of the same problem at Issue #326. --Fred |
Original comment by Pavlos Parissis (Bitbucket: unixsurfer, GitHub: unixsurfer): I have a similar problem where my packages [1] don't show up when I search them using their names. pip install works fine. A google search with 'python haproxytool' returns 1st result the pypi page. [1] anycast-healthchecker, haproxyadmin, haproxystats and haproxytool |
Original comment by Alexey T (Bitbucket: neverwalkaloner, GitHub: neverwalkaloner): Same problem with virgil_sdk package, it doesn't appear in the search results. |
Original comment by Pavlos Parissis (Bitbucket: unixsurfer, GitHub: unixsurfer): Search works fine on http://pydigger.com/ |
Original comment by Michael Howitz (Bitbucket: icemac, GitHub: icemac): @unixsurfer: PyDigger is refreshingly fast but has its own problems. I found the following in the first minutes using it:
In future it could be an alternative but not yet. |
Original comment by Christian Münker (Bitbucket: Chipmuenk, GitHub: Chipmuenk): Underfunding - sometimes I forget that somebody has to pay for the free Bus factor - that is really bad :-( |
Original comment by anatoly techtonik (Bitbucket: techtonik, GitHub: techtonik): You are right. The way to fundraise is through Gratipay (US) or Liberapay (EU), but first it should be discussed how to help. Nadia Eghbal (@nayafia) is researching topic, so reaching out to people first - making it the mafia way - become best friends with developers - may help to at least bootstrap the process. ) Note that bus factor is already in effect, so I guess the best chance is to arrange a meeting at PyCon or EuroPython, but, for example, I can not afford going there even though I have a plenty of time to write here - it is "spending your time" dilemma, you see, and money was invented to solve that conflict. Nobody want to support free beer for "consumers" - it was fun in the old days when people were drinking it together in a real world, but now they grab it and consume alone, and you have no fun dealing with those barbarian people of new age. I guess the root of the problem is that people are "forced" to use open source, not to create it. So fundraising ans support is still a communication/coordination problem, and it requires financial management effort and skills that historically very unusual for open source people, so unless we find the time to invest into communication and sync, we won't see all those additional constraints that are pressing developers these days. |
Original comment by Vaibhav Kohli (Bitbucket: vaibhavkohli91, GitHub: Unknown): I am facing the same issue for my new package Liota, which is getting listed on web search as well as getting installed but not showing up in the pypi search. |
Same problem with dnsdiag package |
For https://pypi.python.org/pypi/testresources package the PyPI search turns up 2 old versions (0.2.7 and 1.0.0) instead of the newest, 2.0.1 |
want to be clear that the pypi maintainers are aware and motivated to work on resolving these concerns. there have just been many more pressing concerns in the maintenance of PyPI and development of the next generation codebase warehouse |
Is there a likely ETA for fixing the search and similar issues, or an ETA for warehouse to replace the existing pypi? I don't think many people know just how broken pypi is, despite it being vitally important to so many projects and services and underpins so much of python! |
What about already using https://pypi.io (aka warehouse)? Its index seems to be complete and correct. For many parts (aka searching a package) it should be sufficient. |
@icemac how to instruct pip to use this as default? |
@farrokhi: Why do you want to use https://pypi.io with |
Its about both, surely... (the pip search is broken too...) |
@icemac they use the same backend afaik |
It seems that there was an update in the search index lately. Yesterday I was able to find the current version of my package named |
Search has been patched up on PyPI. I'm triaging search related issues right now and most of the issues flagged in this thread have been resolved. Please open a new issue with any ongoing search related bugs. |
Originally reported by: Luca Rebuffi (Bitbucket: lucarebuffi, GitHub: lucarebuffi)
I am maintaining packages named "Oasys" and "ShadowOui" in the Pypi site since several months. These packages are available by query with the keyword "oasys".
Since a couple of weeks the result of the query with the keyword and eventually with their names directly does not contain the last versions. On the contrary, all the previous versions, also the ones I deleted are shown. Paradoxically, the command "pip install " finds the correct last version.
The text was updated successfully, but these errors were encountered: