Compilation errors with IFORT 21
まだ誰も着手していません。
評価
- 難易度
- 4/5
- 見積もり時間
- 3〜5日
- 初心者へのやさしさ
- 35/100
- issue の種類
- バグ
- 明瞭さ
- おおむね明確
- 活発さ
- 停滞
- 技術スタック
- fortran
- 領域
- build-system
調査の方向性
Start by reading the compiler-sensitive code in lin_stdlib_blas_constants.f90, stdlib_error.f90 around appendr, and stdlib_linalg_lapack_aux.F90 around stdlib_ieeeck. Reproduce the reported failures with IFORT 21 on Debian 11 and check whether the quick fixes preserve rank-zero handling and IEEE checks. Done means the sources compile successfully with IFORT 21 while retaining the intended behavior.
索引モデルが issue の本文から書いたものです。
説明
Description
I have downloaded the sources from the fpm branch and tried to compile them "manually" (I mean without fpm). The objective is to put the sources in our industrial software, which has its own build system, and IFORT 21 is the compiler I have to use (and which is not the list of the tested compilers).
The compilation is mostly OK, except for a couple of files, for which I had to apply some quick (and dirty?) fixes... Here is the list:
2 modules in a single file lin_stdlib_blas_constants.f90
It contains lin_stdlib_blas_constants_sp and lin_stdlib_blas_constants_dp. Our dependency management assumes that a source file can contain only one module, with the exact same name as the file (without the .f90 suffix). I think it's a very common convention, and it would be nice to split this into two files.
Internal compiler error in stdlib_error.f90
It happens here:
pure subroutine appendr(msg,a,prefix)
class(*),optional,intent(in) :: a(..)
character(len=*),intent(inout) :: msg
character,optional,intent(in) :: prefix
if (present(a)) then
select rank (v=>a)
rank (0) !!! line 383 !!
call append (msg,v,prefix)
rank (1)
call appendv(msg,v)
rank default
msg = trim(msg)//' <ERROR: INVALID RANK>'
end select
endif
end subroutine appendr
stdlib/lin_stdlib_error.f90(383): catastrophic error: **Internal compiler error: internal abort**
I had to rewrite it to make it compile, but obviously nothing will be done for rank(a)==0, and I don't know if it's a problem or not:
pure subroutine appendr(msg,a,prefix)
class(*),optional,intent(in) :: a(..)
character(len=*),intent(inout) :: msg
character,optional,intent(in) :: prefix
if (present(a)) then
if (rank(a) == 0) return
end if
if (present(a)) then
select rank (v=>a)
!rank (0) !!! line 383 !!
! call append (msg,v,prefix)
rank (1)
call appendv(msg,v)
rank default
msg = trim(msg)//' <ERROR: INVALID RANK>'
end select
endif
end subroutine appendr
Unrecognized functions in stdlib_linalg_lapack_aux.F90
It happens here:
pure integer(ilp) function stdlib_ieeeck( ispec, zero, one )
!! IEEECK is called from the ILAENV to verify that Infinity and
!! possibly NaN arithmetic is safe (i.e. will not trap).
! -- lapack auxiliary routine --
! -- lapack is a software package provided by univ. of tennessee, --
! -- univ. of california berkeley, univ. of colorado denver and nag ltd..--
! Scalar Arguments
integer(ilp), intent(in) :: ispec
real(sp), intent(in) :: one, zero
! =====================================================================
! Executable Statements
stdlib_ieeeck = 1
! Test support for infinity values
if (.not.ieee_support_inf(one)) then
stdlib_ieeeck = 0
return
end if
! return if we were only asked to check infinity arithmetic
if (ispec == 0) return
if (.not.ieee_support_nan(one)) then
stdlib_ieeeck = 0
return
end if
return
end function stdlib_ieeeck
stdlib/lin_stdlib_linalg_lapack_aux.F90(138): error #7137: Any procedure referenced in a PURE procedure, including one referenced via a defined operation or assignment, must have an explicit interface and be declared PURE. [IEEE_ARITHMETIC^FOR_IEEE_SUPPORT_INF]
if (.not.ieee_support_inf(one)) then
--------------------^
stdlib/lin_stdlib_linalg_lapack_aux.F90(146): error #7137: Any procedure referenced in a PURE procedure, including one referenced via a defined operation or assignment, must have an explicit interface and be declared PURE. [IEEE_ARITHMETIC^FOR_IEEE_SUPPORT_NAN]
if (.not.ieee_support_nan(one)) then
--------------------^
This one is weird... I ended up by simply returning the value $0$, but again I don't know what are the consequences:
```fortran
pure integer(ilp) function stdlib_ieeeck( ispec, zero, one )
integer(ilp), intent(in) :: ispec
real(sp), intent(in) :: one, zero
stdlib_ieeeck = 0
end function stdlib_ieeeck
Expected Behaviour
successful compilation ;)
Version of stdlib
0.7.0
Platform and Architecture
Debian 11
Additional Information
No response
- 主要言語
- Fortran
- スター
- 1.4k
- フォーク
- 233
- 平均マージ
- 4日 11時間
- マージ済み PR(30日)
- 5
環境構築
- Dockerfile・Docker Compose ファイルなし
- プルリクエストのテンプレートあり
- コントリビューションガイドを読む
はじめの一歩
- issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
- 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
- リポジトリをフォークし、ブランチを切って変更します。
- issue 番号を参照したプルリクエストを送ります。
fortran-lang/stdlib のほかの issue
-
難易度 2/5 1〜3時間 初心者へのやさしさ 68/100
fortran-lang/stdlib#1193 · コメント 1 件 ·
-
bug
難易度 3/5 1〜2日 初心者へのやさしさ 58/100
fortran-lang/stdlib#1224 · コメント 2 件 ·
-
LAPACK upgradesオープンidea topic: linalg
難易度 5/5 1週間以上 初心者へのやさしさ 25/100
fortran-lang/stdlib#1199 · リアクション 1 件 ·
-
idea
難易度 5/5 1週間以上 初心者へのやさしさ 35/100
fortran-lang/stdlib#1181 · コメント 2 件 · リアクション 1 件 ·
-
bug
難易度 2/5 1〜3時間 初心者へのやさしさ 49/100
fortran-lang/stdlib#1178 · コメント 3 件 ·
fortran-lang/stdlib の issue をすべて見る
似ている issue
-
fix(schema-drift): make launcher-manpage fail on mandoc errors and anchor secret env-var matchesオープンagent-butler-finding bug
難易度 2/5 1〜3時間 初心者へのやさしさ 82/100
jordansmall/spindrift#6160 ·
メンテナーはふだん 1 日以内に返信
-
nextjs-tracking
難易度 2/5 1〜3時間 初心者へのやさしさ 67/100
cloudflare/vinext#4390 ·
メンテナーはふだん 1 日以内に返信
-
難易度 1/5 1〜3時間 初心者へのやさしさ 90/100
refactoringhq/tolaria#1303 ·
メンテナーはふだん 1 日以内に返信
-
Primary JSON and TTL release artefacts get wrong ontology IRI (ONTBASE instead of URIBASE)対応中かも @matentzn が今日担当しました。 オープン
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
メンテナーはふだん 1 日以内に返信
-
難易度 2/5 1〜3時間 初心者へのやさしさ 72/100