Deindex several ports that were =skip. #42222

BillyONeal · 2024-11-19T01:09:07Z

Removes several ports from the registry that conflict with others and/or are effectively untested in vcpkg's build lab.

[cppmicroservices] This port was added 5 years ago, appears relatively unused, only exists in repology due to vcpkg, and is full of vendored dependencies. It's also broken for most official triplets. Replaces #34552
[ijg-libjpeg, mozjpeg] Leaf ports that we can't test and conflict with libjpeg-turbo, the curated registry's canonical jpeg implementation.
[optional-bare] Conflicts with optional-lite.
[qt5-canvas3d] Removed by qt5 upstream.
[range-v3-vs2015] No longer maintained by upstream, no longer necessary in 2024.
[rbdl-orb] File conflicts with its own dependency.
[sciter] Left support by upstream 2 years ago and has conflicts with other ports.
[signalrclient] "Obsolete" and conflicts with microsoft-signalr
[spdk-*] Several conflicts with other ports.
[c-dbg-macro] Conflicts with [dbg-macro]
[casadi] #41649 appears to have been merged in error, casadi/casadi#3896 needs to land before this can be in the curated registry. Moreover, problems in the portfile were skipped in ci.baseline.txt rather than a supports expression. We are happy to bring this port back as soon as that issue is fixed :) [chromium-base] Not tested anywhere for years.
[libmesh] Conflicts with other ports.
[libpng-apng] Doesn't even exist in the first place.
[libqcow] Release distribution file contains configured sources, while the source code in the repository does not.
[libvmdk] Upstream is gone.
[mecab] Upstream has not been edited in 10 years, combination of =skip and supports clause means not tested.
[shogun] Not tested.

Not deindexed:
[bde] Conflicts with other ports. Warning was given more than a year ago: #32645 . However, #41335 appears to indicate that it works now, so the =skips were removed.
[stxxl] Appears to work on my machine?
[chartdir] Unstable download URLs are now mitigated by asset caching.
[buck-yeh-bux] Works on my machine or already skipped by supports clause. The mariadb backend is broken, but it's entirely broken. We should consider deindexing completely, but @buck-yeh appears to be an active contributor here so I want to give them the heads' up before doing anything with it.
[boringssl, fluidlite, libressl, libmariadb, libmysql] Appears to have a dedicated userbase despite being untestable in CI.
[libxaw] Needed supports expression for ONLY_DYNAMIC_LIBRARY, should have always been a fail not a skip.

ntf-core:arm64-osx=fail
ntf-core:x64-linux=fail
ntf-core:x64-osx=fail
ntf-core:x64-windows-static-md=fail
ntf-core:x64-windows-static=fail
ntf-core:x64-windows=fail
ntf-core:x86-windows=fail

These were previously 'hidden' by bde=skip

Related: #42165

Removes several ports from the registry that conflict with others and/or are effectively untested in vcpkg's build lab. [cppmicroservices] This port was added 5 years ago, appears relatively unused, only exists in repology due to vcpkg, and is full of vendored dependencies. It's also broken for most official triplets. Replaces microsoft#34552 [ijg-libjpeg, mozjpeg] Leaf ports that we can't test and conflict with libjpeg-turbo, the curated registry's canonical jpeg implementation. [optional-bare] Conflicted with optional-lite. [qt5-canvas3d] Removed by qt5 upstream. [range-v3-vs2015] No longer maintained by upstream, no longer necessary in 2024. [rbdl-orb] File conflicts with its own dependency. [sciter] Left support by upstream 2 years ago and has conflicts with other ports. [signalrclient] "Obsolete" and conflicts with microsoft-signalr [spdk-*] Several conflicts with other ports. [c-dbg-macro] Conflicts with [dbg-macro] [casadi] microsoft#41649 appears to have been merged in error, casadi/casadi#3896 needs to land before this can be in the curated registry. Moreover, problems in the portfile were skipped in ci.baseline.txt rather than a supports expression. We are happy to bring this port back as soon as that issue is fixed :) [chromium-base] Not tested anywhere for years. [libmesh] Conflicts with other ports. [libpng-apng] Doesn't even exist in the first place. [libqcow] Release distribution file contains configured sources, while the source code in the repository does not. [libvmdk] Upstream is gone. [mecab] Upstream has not been edited in 10 years, combination of =skip and supports clause means not tested. [shogun] Not tested. Not deindexed: [bde] Conflicts with other ports. Warning was given more than a year ago: microsoft#32645 . However, microsoft#41335 appears to indicate that it works now, so the =skips were removed. [stxxl] Appears to work on my machine? [chartdir] Unstable download URLs are now mitigated by asset caching. [buck-yeh-bux] Looks like these are intended to be in the supports clause. [boringssl, fluidlite, libressl, libmariadb, libmysql] Appears to have a dedicated userbase despite being untestable in CI. [libxaw] Needed supports expression for ONLY_DYNAMIC_LIBRARY, should have always been a fail not a skip. Related: microsoft#42165

dg0yt · 2024-11-19T05:00:15Z

[boringssl, fluidlite, libressl, libmariadb, libmysql] Appears to have a dedicated userbase despite being untestable in CI.

libmariadb builds in CI on android.
It was meant to build in x86-windows, libmysql everywhere else.
But both the x86-windows and the android results are ... unexpected.
(By AZP tests analytics.)

Add ntf-core, previously hidden by skipped bde.

BillyONeal · 2024-11-19T05:13:07Z

libmariadb builds in CI on android.

True, we have that funny arrangement for only libmariadb/libmysql.

BillyONeal · 2024-11-19T05:17:02Z

scripts/ci.baseline.txt

-catch-classic:x64-windows        = skip
-catch-classic:x64-windows-static = skip
-catch-classic:x64-windows-static-md=skip
-catch-classic:x86-windows        = skip
 cctag:x64-windows-static-md=fail
 cello:arm64-uwp=fail
 cello:x64-uwp=fail
 chakracore:x64-linux=skip
 # chartdir does not offer stable download URLs


Suggested change

# chartdir does not offer stable download URLs

BillyONeal · 2024-11-19T06:22:30Z

Tagging vcpkg-team-review because

I'm the author and need another team member to OK
Deindexing generally requires more than 1 maintainer to OK

dg0yt

Successful builds in CI.
(Not necessarily indicating usefullness.)

dg0yt · 2024-11-19T06:53:03Z

versions/baseline.json

@@ -1460,10 +1460,6 @@
      "baseline": "1.34.2",
      "port-version": 0
    },
-    "c-dbg-macro": {


CI on android (header-only). But probably a bad idea, as on other platforms.

dg0yt · 2024-11-19T06:56:00Z

versions/baseline.json

@@ -1520,10 +1516,6 @@
      "baseline": "1.0.0",
      "port-version": 6
    },
-    "casadi": {


CI on linux, osx, windows-dynamic.

dg0yt · 2024-11-19T06:58:22Z

versions/baseline.json

@@ -1980,10 +1964,6 @@
      "baseline": "0.4.0",
      "port-version": 0
    },
-    "cppmicroservices": {


CI on 6/12 triplets. (Related: #34552.)

This one has CI, but vendored dependencies that stomp on other things. (In fact this is the port that got me to check all the =skip ones in the first place)

dg0yt · 2024-11-19T07:00:13Z

versions/baseline.json

@@ -5004,10 +4976,6 @@
      "baseline": "0.21.5",
      "port-version": 1
    },
-    "libqcow": {


CI on arm64-osx. But known bad. (Related: #36809.)

dg0yt · 2024-11-19T07:01:40Z

versions/baseline.json

@@ -5872,10 +5836,6 @@
      "baseline": "0.6.0",
      "port-version": 0
    },
-    "mecab": {


CI on android, osx.

Hmmmm.... I may have been to hasty with this one

dg0yt · 2024-11-19T07:06:16Z

versions/baseline.json

@@ -8348,10 +8284,6 @@
      "baseline": "1.0",
      "port-version": 7
    },
-    "shogun": {


CI on linux.

dg0yt · 2024-11-19T07:07:46Z

versions/baseline.json

-      "baseline": "24.01",
-      "port-version": 0
-    },
-    "spdk-dpdk": {


CI on linux.

The spdk-* ones have CI, but appear to be effectively vendored and modified copies of other things in the registry.

dg0yt · 2024-11-19T07:08:00Z

versions/baseline.json

-      "baseline": "2018-11-24",
-      "port-version": 3
-    },
-    "spdk-ipsec": {


CI on linux.

dg0yt · 2024-11-19T07:08:14Z

versions/baseline.json

-      "baseline": "2018-07-11",
-      "port-version": 3
-    },
-    "spdk-isal": {


CI on arm64-osx.

autoantwort · 2024-11-19T12:02:49Z

You could use microsoft/vcpkg-tool#802 to test ports in ci even if they conflict.

BillyONeal · 2024-11-19T18:46:59Z

Unfortunately we don't really have the cores available to be adding multiple 'ways' of testing the registry as the pool already gets very behind a lot of the time. And the combinatoric explosion quickly becomes infeasible.

BillyONeal · 2024-11-22T23:38:33Z

Yesterday @ras0219-msft, @vicroms, @JavierMatosD and I discussed this and agree that only things which are entirely untested on all platforms should be deindexed in a change like this, under the rationale that we would not add such ports today. I will keep the ones dg0yt noted above and split any remaining ones I think should be removed for other reasons into their own PRs

BillyONeal added info:internal This PR or Issue was filed by the vcpkg team. category:infrastructure Pertaining to the CI/Testing infrastrucutre labels Nov 19, 2024

JonLiu1993 assigned Cheney-W Nov 19, 2024

Restore buck-yeh-bux-mariadb-client:x86-windows.

f400ef7

Add ntf-core, previously hidden by skipped bde.

BillyONeal commented Nov 19, 2024

View reviewed changes

BillyONeal added the requires:vcpkg-team-review This PR or issue requires someone on the vcpkg team to take a further look. label Nov 19, 2024

dg0yt reviewed Nov 19, 2024

View reviewed changes

BillyONeal marked this pull request as draft November 20, 2024 19:45

Cheney-W removed the requires:vcpkg-team-review This PR or issue requires someone on the vcpkg team to take a further look. label Nov 22, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Deindex several ports that were =skip. #42222

Deindex several ports that were =skip. #42222

BillyONeal commented Nov 19, 2024 •

edited

Loading

dg0yt commented Nov 19, 2024

BillyONeal commented Nov 19, 2024

BillyONeal Nov 19, 2024

BillyONeal commented Nov 19, 2024

dg0yt left a comment

dg0yt Nov 19, 2024

dg0yt Nov 19, 2024

dg0yt Nov 19, 2024

BillyONeal Nov 19, 2024

dg0yt Nov 19, 2024

dg0yt Nov 19, 2024

BillyONeal Nov 19, 2024

dg0yt Nov 19, 2024

dg0yt Nov 19, 2024

BillyONeal Nov 19, 2024

dg0yt Nov 19, 2024

dg0yt Nov 19, 2024

autoantwort commented Nov 19, 2024

BillyONeal commented Nov 19, 2024 •

edited

Loading

BillyONeal commented Nov 22, 2024

Deindex several ports that were =skip. #42222

Are you sure you want to change the base?

Deindex several ports that were =skip. #42222

Conversation

BillyONeal commented Nov 19, 2024 • edited Loading

dg0yt commented Nov 19, 2024

BillyONeal commented Nov 19, 2024

Choose a reason for hiding this comment

BillyONeal commented Nov 19, 2024

dg0yt left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

autoantwort commented Nov 19, 2024

BillyONeal commented Nov 19, 2024 • edited Loading

BillyONeal commented Nov 22, 2024

BillyONeal commented Nov 19, 2024 •

edited

Loading

BillyONeal commented Nov 19, 2024 •

edited

Loading