Tarantool development patches archive
 help / color / mirror / Atom feed
From: Sergey Kaplun via Tarantool-patches <tarantool-patches@dev.tarantool.org>
To: Maxim Kokryashkin <max.kokryashkin@gmail.com>
Cc: tarantool-patches@dev.tarantool.org
Subject: Re: [Tarantool-patches] [PATCH luajit v2] test: adapt tests checking debug line hook trace
Date: Thu, 16 Dec 2021 18:24:46 +0300	[thread overview]
Message-ID: <YbtaPskLnL45+dLe@root> (raw)
In-Reply-To: <20211110192042.1687467-1-m.kokryashkin@tarantool.org>

Hi, Maxim!
Thanks for the fixes!

LGTM, except a few nits below.

Am I right that this CI branch [1] is the latest one?

Also, please align comments with linewidth 66 symbols.

On 10.11.21, Maxim Kokryashkin wrote:
> The LuaJIT's virtual machine interprets the bytecode
> following the return from function (i.e. the one succeeding
> the call made) and located on the line other than that return
> bytecode, as a new line trigger for line hooks, unlike Lua
> does.
> Here is an example (it is joined in one line intend):
> 
> debug.sethook(function(_, l) print("LINE: "..l) end, "l") loadstring("\n\ns=nil")() debug.sethook()
> 
> This chunk prints for LuaJIT:
> 
> LINE: 3
> LINE: 1
> 
> But for Lua 5.1 it is only "LINE: 3" in the output.
> See also tarantool/tarantool#5693.

Nit: this see also is excess (as far as this patch resolves the ticket).

> 
> Those tests are adapted to LuaJIT behavior by changing assertions
> considering line events that LuaJIT generates.
> 
> Part of tarantool/tarantool#5870

It's OK to mention that this patch "Resolves tarantool/tarantool#5693".

> ---
> Changes in v2:
> - Fixed comments as per review by Sergey
> - Added some comments
> 
>  test/PUC-Rio-Lua-5.1-tests/db.lua | 57 +++++++++++++++++--------------
>  1 file changed, 31 insertions(+), 26 deletions(-)
> 
> diff --git a/test/PUC-Rio-Lua-5.1-tests/db.lua b/test/PUC-Rio-Lua-5.1-tests/db.lua
> index 56f59ea8..120fc04c 100644
> --- a/test/PUC-Rio-Lua-5.1-tests/db.lua
> +++ b/test/PUC-Rio-Lua-5.1-tests/db.lua
> @@ -8,6 +8,23 @@ do
>  local a=1
>  end
> 
> +-- The LuaJIT's virtual machine interprets the bytecode
> +-- following the return from function (i.e. the one succeeding
> +-- the call made) and located on the line other than that return
> +-- bytecode, as a new line trigger for line hooks, unlike Lua
> +-- does.
> +-- Here is an example (it is joined in one line intend):
> +--[[
> +debug.sethook(function(_, l) print("LINE: "..l) end, "l") loadstring("\n\ns=nil")() debug.sethook()
> +--]]
> +-- This chunk prints for LuaJIT:
> +--[[
> +LINE: 3
> +LINE: 1
> +--]]
> +-- But for Lua 5.1 it is only "LINE: 3" in the output.
> +-- See also https://github.com/tarantool/tarantool/issues/5693.
> +-- This function is modified to correspond with LuaJIT's line triggers.
>  function test (s, l, p)
>    collectgarbage()   -- avoid gc during trace
>    local function f (event, line)
> @@ -16,6 +33,11 @@ function test (s, l, p)
>      if p then print(l, line) end
>      assert(l == line, "wrong trace!!")
>    end
> +  -- Despite `loadstring` and `debug.sethook` are on the same line, LuaJIT
> +  -- generates separate line events for them. The test is adapted to LuaJIT
> +  -- behavior by adding corresponding line numbers into the test table.
> +  table.insert(l, 1, 41)
> +  table.insert(l, 41)
>    debug.sethook(f,"l"); loadstring(s)(); debug.sethook()
>    assert(table.getn(l) == 0)
>  end
> @@ -25,8 +47,8 @@ do
>    local a = debug.getinfo(print)
>    assert(a.what == "C" and a.short_src == "[C]")
>    local b = debug.getinfo(test, "SfL")
> -  assert(b.name == nil and b.what == "Lua" and b.linedefined == 11 and
> -         b.lastlinedefined == b.linedefined + 10 and
> +  assert(b.name == nil and b.what == "Lua" and b.linedefined == 28 and
> +         b.lastlinedefined == b.linedefined + 15 and
>           b.func == test and not string.find(b.short_src, "%["))
>    assert(b.activelines[b.linedefined + 1] and
>           b.activelines[b.lastlinedefined])
> @@ -95,26 +117,6 @@ repeat
>    assert(g(f) == 'a')
>  until 1
> 
> --- FIXME: The LuaJIT's virtual machine interprets the bytecode
> --- following the return from function (i.e. the one succeeding
> --- the call made) and located on the line other than that return
> --- bytecode, as a new line trigger for line hooks, unlike Lua
> --- does.
> --- Here is an example (it is joined in one line intend):
> ---[[
> -debug.sethook(function(_, l) print("LINE: "..l) end, "l") loadstring("\n\ns=nil")() debug.sethook()
> ---]]
> --- This chunk prints for LuaJIT:
> ---[[
> -LINE: 3
> -LINE: 1
> ---]]
> --- But for Lua 5.1 it is only "LINE: 3" in the output.
> --- See also https://github.com/tarantool/tarantool/issues/5693.
> --- Considering implementation-defined behaviour difference
> --- (see also https://luajit.org/status.html) test is disabled for
> --- LuaJIT.
> ---[=[
>  test([[if
>  math.sin(1)
>  then
> @@ -132,23 +134,25 @@ else
>  end
>  ]], {2,5,6})
> 
> +-- Test is adapted to the behaviour of LuaJIT.
>  test([[a=1
>  repeat
>    a=a+1
>  until a==3
> -]], {1,3,4,3,4})
> +]], {1,2,3,4,2,3,4})
> 
>  test([[ do
>    return
>  end
>  ]], {2})
> 
> +-- Test is adapted to the behaviour of LuaJIT.
>  test([[local a
>  a=1
>  while a<=3 do
>    a=a+1
>  end
> -]], {2,3,4,3,4,3,4,3,5})
> +]], {1,2,3,4,3,4,3,4,3,5})
> 
>  test([[while math.sin(1) do
>    if math.sin(1)
> @@ -168,8 +172,9 @@ test([[for i,v in pairs{'a','b'} do
>  end
>  ]], {1,2,1,2,1,3})
> 
> -test([[for i=1,4 do a=1 end]], {1,1,1,1,1})
> ---]=]
> +-- Test is adapted to the behaviour of LuaJIT, as it generates only four line
> +-- events, unlike Lua, which generates five of them.
> +test([[for i=1,4 do a=1 end]], {1,1,1,1})
> 
> 
>  print'+'
> --
> 2.33.0
> 

[1]:
https://github.com/tarantool/tarantool/tree/fckxorg/gh-5693-PUC-Rio-debug-line-hook

-- 
Best regards,
Sergey Kaplun

  reply	other threads:[~2021-12-16 15:26 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-11-10 19:20 Maxim Kokryashkin via Tarantool-patches
2021-12-16 15:24 ` Sergey Kaplun via Tarantool-patches [this message]
2022-01-24 22:02   ` Максим Корякшин via Tarantool-patches

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=YbtaPskLnL45+dLe@root \
    --to=tarantool-patches@dev.tarantool.org \
    --cc=max.kokryashkin@gmail.com \
    --cc=skaplun@tarantool.org \
    --subject='Re: [Tarantool-patches] [PATCH luajit v2] test: adapt tests checking debug line hook trace' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox