Skip to content

Make the directory layout match upstream MSYS #16

Closed
Closed
Make the directory layout match upstream MSYS#16
Assignees
sschuberth
Labels
sdk
Milestone
First release of Git for Windows generated using the new SDK

Description

@sschuberth
sschuberth
opened on Oct 15, 2014

Currently we have "swapped" the MSYS and MinGW directory trees via a profile.xml to match what we had for msysgit. However, it makes more sense for us to simplify things by aligning the directory layout again with upstream MSYS.

Activity

  1. sschuberth
    added
    sdk
    on Oct 15, 2014
  2. sschuberth
    self-assigned this
    on Oct 15, 2014
  3. sschuberth
    added this to the First release of Git for Windows generated using the new SDK milestone on Oct 15, 2014

  4. dscho commented on Oct 15, 2014

    @dscho
    dscho
    on Oct 15, 2014
    Member

    👍

  5. dscho commented on Oct 15, 2014

    @dscho
    dscho
    on Oct 15, 2014
    Member

    It would be enough for that profile.xml to lose its <system-map> section, yes? I.e.

    <?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<profile project="MinGW" application="mingw-get">
  <repository uri="http://prdownloads.sourceforge.net/mingw/%F.xml.lzma?download">
  </repository>
  <repository uri="https://github.com/git-for-windows/sdk-packages/blob/master/%F.xml.lzma?raw=true">
    <package-list catalogue="git-win-sdk-package-list" />
  </repository>
  <system-map id="default">
    <sysroot subsystem="mingw32" path="%R" />
    <sysroot subsystem="msys" path="%R/../" />
  </system-map>
</profile>

    would become

    <?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<profile project="MinGW" application="mingw-get">
  <repository uri="http://prdownloads.sourceforge.net/mingw/%F.xml.lzma?download">
  </repository>
  <repository uri="https://github.com/git-for-windows/sdk-packages/blob/master/%F.xml.lzma?raw=true">
    <package-list catalogue="git-win-sdk-package-list" />
  </repository>
</profile>

    ... and then we have to identify all the places that still expect the msysGit layout... ;-)

  6. sschuberth commented on Oct 15, 2014

    @sschuberth
    sschuberth
    on Oct 15, 2014
    Author

    Exactly. Your last sentence names the hard part.

  7. t-b commented on Oct 15, 2014

    @t-b
    t-b
    on Oct 15, 2014

    Does that mean we have to tweak every mgwport file?

  8. sschuberth commented on Oct 15, 2014

    @sschuberth
    sschuberth
    on Oct 15, 2014
    Author

    Probably not tweak every file, but we must ensure for every file that it still builds, and if not, then tweak it.

  9. t-b commented on Oct 15, 2014

    @t-b
    t-b
    on Oct 15, 2014

    Just to understand the implications. In the curl mgwport file I would need to change

    insinto /mingw/include/curl

    to

    insinto /include/curl

    or?

    And while we are at it, it might also make sense to use the generic commands like doinclude instead of the ones hardcoding the paths with insinto.

  10. sschuberth commented on Oct 15, 2014

    @sschuberth
    sschuberth
    on Oct 15, 2014
    Author

    Correct.

  11. t-b commented on Oct 15, 2014

    @t-b
    t-b
    on Oct 15, 2014

    And how are $__mpflavor_prefix and friends from mgwport effected?
    Or to rephrase my question: Can we write mgwport files which are before and after the switch valid?

  12. sschuberth commented on Oct 15, 2014

    @sschuberth
    sschuberth
    on Oct 15, 2014
    Author

    Can we write mgwport files which are before and after the switch valid?

    Probably yes, I don't know and would have to research.

  13. dscho commented on Oct 15, 2014

    @dscho
    dscho
    on Oct 15, 2014
    Member

    @sschuberth do you think we will have to switch all #!/bin/sh to #!/msys/1.0/bin/sh with this switch? If so, I think we should not switch.

  14. sschuberth commented on Oct 15, 2014

    @sschuberth
    sschuberth
    on Oct 15, 2014
    Author

    No, I'm very sure we don't need to do that, as Shell scripts that come from upstream also don't have that.

  15. sschuberth commented on Oct 23, 2014

    @sschuberth
    sschuberth
    on Oct 23, 2014
    Author

    Hmm, maybe we need to re-think this. In upstream MinGW / MSYS (installed to C:\MinGW), if you start C:\MinGW\msys\1.0\msys.bat, you get this:

    $ cd /
$ pwd -W
C:/MinGW/msys/1.0

    I.e., a cd / does not take you to the installation directory, but to the root of the MSYS directory. While this makes sense from an MSYS point of view, it makes it unnecessary hard to access files in C:\MinGW from the MSYS shell. So, what do you think, should we keep things as-is?

  16. 59 remaining items

  17. mjcheetham
    added a commit that references this issue on Mar 26, 2025

    Fix rare segfault in sparse-index (git-for-windows#690)

    281fa06
  18. dscho
    added a commit that references this issue on Aug 13, 2025

    Fix rare segfault in sparse-index (#690)

    43e5af3
  19. dscho
    added 2 commits that reference this issue on Dec 9, 2025

    builtin/gc: fix condition for whether to write commit graphs

    4029162

    odb: properly close sources before freeing them

    e80c43d
  20. git-for-windows-ci
    added 2 commits that reference this issue on Dec 11, 2025

    builtin/gc: fix condition for whether to write commit graphs

    fc98532

    odb: properly close sources before freeing them

    8915881
  21. mjcheetham
    added a commit that references this issue on Dec 17, 2025

    Fix rare segfault in sparse-index (git-for-windows#690)

    c64ba36
  22. git-for-windows-ci
    added 2 commits that reference this issue on Jan 8, 2026

    builtin/gc: fix condition for whether to write commit graphs

    b3449b1

    odb: properly close sources before freeing them

    3d09968
  23. dscho
    added a commit that references this issue on Jan 29, 2026

    Fix rare segfault in sparse-index (#690)

    070c66e
  24. gitforwindowshelper
    added 4 commits that reference this issue on Jun 25, 2026

    reftable/table: fix NULL pointer access when seeking to bogus offsets

    aec1866

    reftable/table: fix NULL pointer access when seeking to bogus offsets

    83b2233

    reftable/block: fix OOB read with bogus block size

    43a0669

    reftable/table: fix NULL pointer access when seeking to bogus offsets

    a1c085d
  25. gitforwindowshelper
    added a commit that references this issue on Sep 14, 2026

    builtin/history: unuse the commit buffer after use

    52a351e
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

sdk

Type

No type

Projects

No projects

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions