Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 06:52
    ozbenh edited #2205
  • 06:51
    ozbenh opened #2205
  • Oct 03 18:36
    Aniruddha-Deb commented #2203
  • Oct 03 17:01

    github-actions[bot] on nightly

    synth: avoid crash on invalid h… testsuite/synth: add a test for… (compare)

  • Oct 03 15:08
    tgingold closed #2204
  • Oct 03 15:08

    tgingold on master

    synth: avoid crash on invalid h… testsuite/synth: add a test for… (compare)

  • Oct 03 14:27
    ekliptik commented #2204
  • Oct 03 14:10
    ekliptik edited #2204
  • Oct 03 14:10
    ekliptik edited #2204
  • Oct 03 14:08
    ekliptik opened #2204
  • Oct 03 12:37
    tgingold commented #2203
  • Oct 03 10:55

    github-actions[bot] on nightly

    workflows: tentatively build on… workflow: use llvm-14 on unbunt… workflow: back to ubuntu 18. R… and 1 more (compare)

  • Oct 03 09:50

    tgingold on master

    gh action: replace ubuntu 18 by… (compare)

  • Oct 03 08:40

    tgingold on gha

    workflow: test only seletected … (compare)

  • Oct 03 08:28

    tgingold on gha

    workflow: add sudo (compare)

  • Oct 03 08:15

    tgingold on gha

    gha: install gnat in the non-do… (compare)

  • Oct 03 07:37

    tgingold on action

    (compare)

  • Oct 03 07:37

    tgingold on gha

    workflow: do not use docker for… (compare)

  • Oct 03 06:50

    tgingold on master

    workflow: use llvm-14 on unbunt… workflow: back to ubuntu 18. R… (compare)

  • Oct 02 18:16

    tgingold on issue1372

    (compare)

Ed Bordin
@edbordin
I was calling make with GNAT_BARGS="-bargs -E -static" GNAT_LARGS="-static -lz" to force a static link with zlib, is there a better way to do it? Perhaps I should just open a GH issue
in the meantime I might just patch it back in to get things building again
tgingold
@tgingold
GNAT_BARGS still exists, but GNAT_LARGS has been replaced by LDFLAGS.
Ed Bordin
@edbordin
LDFLAGS didn't seem to quite behave the same
the linker flags I'm adding are very sensitive to ordering though because I'm adding -static so I was possibly never using it as intended
tgingold
@tgingold
Do not hesitate to propose a change!
Ed Bordin
@edbordin
I've run out of time today but yes, I agree I should collaborate with you rather than patching downstream!
it doesn't help that I don't really understand all the details of gnatlink etc.
xiretza
@xiretza:xiretza.xyz
[m]
@edbordin: that one's on me, sorry about that, we could just revert that commit
or add a generic GNAT_FLAGS variable
Unai Martinez-Corral
@umarcor
I will rebase ghdl/ghdl#1547, so we can use it as a reference for the windows builds. That covers all the working and non-working setups I could gather.
tgingold
@tgingold
@edbordin For gnatlink, just use -v so that it displays how gcc is called. Then you need to narrow down which command is not correct and what would be the best order for the options.
Ed Bordin
@edbordin
It looks like the fpga-toolchain static builds will soon be superseded by ones managed by yosyshq, so I might just leave it as-is
(I knew this was happening but wasn't sure what the timeline was)
@xiretza:xiretza.xyz yeah that's basically what I ended up doing with a patch, works fine for now. If nobody else is using that flag it probably doesn't need to be reverted
T. Meissner
@tmeissner
Not so cool - GHDL is not included.
T. Meissner
@tmeissner
Time for PRs :smile:
Ed Bordin
@edbordin
they're working on it I think, it's just complicated because it's all cross-compiled from linux
T. Meissner
@tmeissner
Ah, cool :smiley:
Unai Martinez-Corral
@umarcor
I really hope the fpga-toolchain is kept until the oss-cad-suite is improved. Currently, they are reinventing appimage/flatpak manually...
There is 1GB of Ubuntu system libs included in the tarball, and all the executables are wrapped in bash/perl scripts for overriding the linker and library paths.
Martin
@hackfin
Oh my god. Sounds like the Xilinx method.
Unai Martinez-Corral
@umarcor
It is...
Martin
@hackfin
Well, someone will go and dockerize it, eventually.
Unai Martinez-Corral
@umarcor
Most of the tools are dockerised already...
So, fpga-toolchain makes sense because it's a different (all static) solution.
In oss-cad-suite, they are partially duplicating the MINGW packages and they are partially duplicating the hdl/containers.
See YosysHQ/oss-cad-suite-build#1.
Nonetheless, if they want to provide "their own solution for their clients" it makes sense for them to have a branded bundle.
Unai Martinez-Corral
@umarcor
I was about to ask something in the gitter channel of another open source community and I saw they moved to https://zulip.com a week ago.
I didn't know about that software/service.
It looks really nice. It's open source. The default free plan is limited to 10k messages of search history and 5GB of storage. However, they provide the standard plan for open source projects: https://zulip.com/for/open-source/.
Moreover, open source projects can open invitations, so that anyone can join without one.
We have previously discussed how awful the threading solution in gitter is, and how barely usable it is on smartphones.
I believe the main reason for staying here is the very nice integration with github/gitlab for cross-references.
Zulip seems to have a nice integration too, and users can also login using github/gitlab accounts.
Unai Martinez-Corral
@umarcor
The main drawback I see is that zulip communities seem not to be public. So, users can join freely if we enable it, but they do need to join in order to read the content. I really like that gitter allows users to reads without requiring them to register and login.
Any experiences/thoughts?
1 reply
tgingold
@tgingold
No strong opinion. I prefer an open-source solution, but moving has also a cost.
nobodywasishere
@nobodywasishere:eowyn.net
[m]
It could also be that all gitter chatrooms are automatically converted into matrix rooms in the future, in which case no change would be necessary
T. Meissner
@tmeissner
We use an zulip with an internal instance at work. It’s okay, you can have a lot of streams, it supports threads, private communication etc. What I don’t like are their electron based „native“ apps. I use it only in the browser instead.
Unai Martinez-Corral
@umarcor
@nobodywasishere:eowyn.net matrix being a decentralized/federated solution, it can be used regardless of the preferred centralized entrypoint. I don't see the point of making matrix the main entrypoint, because it is an aggregator from a conceptual point of view. Am I misunderstanding it?
nobodywasishere
@nobodywasishere:eowyn.net
[m]
@umarcor I'm not sure what you mean
Unai Martinez-Corral
@umarcor
@nobodywasishere:eowyn.net I mean there is no "matrix client". Matrix is a thing. Clients are a different thing. Therefore, zulip cannot be compared to matrix. zulip/gitter need to be compared to some client of matrix. From the user interaction point of view, having matrix below is not relevant.
That is, regardless of using zulip, gitter or element, matrix can still be used and any user can pick a different client.
My concern with using matrix "only" is that I'm unsure about the matrix protocol supporting non-message events, such as the webhooks from github/gitlab repos. With gitter/zulip, we get those features and users of matrix get the filtered content according to the features in that network.
nobodywasishere
@nobodywasishere:eowyn.net
[m]
From a user point of view, that makes sense.
Unai Martinez-Corral
@umarcor
Nonetheless, I might be confusing matrix and IRC. Maybe matrix does support threads and events.
nobodywasishere
@nobodywasishere:eowyn.net
[m]
To be honest I have no idea
Not even sure what those are used for currently (here)