Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
@nim-nim :
One shouldn't expect `nil == false` any more than one would expect `"yes" ==
true`, although both sides *evaluate to* true, ie `("yes" and true)` *is* true.
The language documentation is quite explicit about this, and this has
wonderfully little to do with rpm or the interface it provides.
Now, there are no clear concise examples provided, so I can only assume some of
this confusion comes from expecting rpm.expand() on conditionals to evaluate
somehow specially, but such evaluation *is not possible*. Rpm macro expansion
is a *string processor* which simply produces a string of zero or more
characters according to its own rules (think printf() on steroids), any
interpretation of the returned string is solely up to the user. If you expect
something else and additionally have misunderstandings about Lua language
fundamentals, I'm sure it's going to be a painful experience. I could've
clarified this anytime if you had just asked. Also, adding separate API to
return a nice little boolean whether a macro is defined or not was a matter of
like 15min of work. All these years and nobody thought to ask for such a simple
thing!
I was hoping for some constructive feedback on the newly added things that are
intended to help that very case, including discussion about usage scenarios
with the specific Lua semantics in mind, but instead I get the same old whining
about how everything is awful (which sure gets tiresome after all these years),
quite clearly without so much as a glance to look at what's being done to
improve things.
I'm not particularly proud of the blown fuse on Friday, but not everything is
rpm's fault.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-707076354___
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
> What I objected here is exposing false instead of nil when something is unset. But there's no such thing in the PR linked to this item at all! That's what's so frustrating to me here and why I've been urging you to look at what was actually implemented for this ticket: https://github.com/rpm-software-management/rpm/commit/67abf72ef57e58251271f5b218f867357d78a4e0 "opts" table always exists so you don't need to test *that* against nil each and every time, and beyond that you get nil's for undefined options, and strings for defined (empty vs non-empty depending on whether it takes an argument or not). The other part is "arg" table (also always existing) where you get leftover arguments as a native Lua sequence, accessible with arg[n] and count with #arg. Other improvements in this area are in the works and some already done, but that's beyond the scope here. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-707141799___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
BTW, the whole state of things is painful and frustrating to me too. If things can not work out today let’s just forget about it for now. I’m not looking for a fight and I have no energy left for arguing in circles. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-707126641___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
@pmatilai Look, the only thing I care about lua side is being able to access spec state easily and reliably. state = state of macros/variables, state of flags and option passed to a macro, etc. The best macro logic in the world will produce erroneous results if it is mistaken about spec state. Right now a lot of this accessing is done via expands. While expand is only a string processor that’s the only tool provided in the documented rpm lua API to read state lua-side. And you often need to compare several slightly different expand calls to guess the real data state. Good replacements to this expand mess should expose real state lua side. Real means nil if unset, false if set to false, true if set to true (or just set without value), empty string if set to empty string, array of values if a macro evaluates to several lines of values (à la sourcelist). While nil and false behave the same in bare conditionals, and nil and empty string behave the same when expanding to string, they are not the same semantically. Typically, nil means you are free to use default values instead, while an explicit false or empty string is an explicit packager request. When a macro can accept both quiet and verbose flags, or a general default state variable, you really really want to know which one of those was actually set to something (including false) and which one of those is currently unset.. What I objected here is exposing false instead of nil when something is unset. Requiring a separate call to check if it’s false or unset is no better than comparing several expand calls (with our without leading ?) to get the actual result. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-707105106___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
Closed #1092 via 67abf72ef57e58251271f5b218f867357d78a4e0. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#event-3859368773___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
Sigh. *Obviously* nil is not *equal to* false, any more than, say, None and False in Python. It would be one bizarre language if they were! Point out concrete problems in the pull-request, or shut up. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-706078696___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
> Oh, looking closer:
>
> > @nim-nim wrote
> > [...]
> > exist because nil is not the same as false or empty
>
> Yes, nil is not same as empty. It however very explicitly is same as false,
> quoting https://www.lua.org/manual/5.4/manual.html#2.1:
(lua)
> if (nil == false) then print("nil is the same thing as false") else
> print("nil and false are different") end
nil and false are different
As this trivial example shows they are most definitely not the same at the
language level, and if you expect them to you **will** get unexpected
comparison and conditional behaviors. (other things that will fail unexpectedly
is transformation to string values, which is where the empty string comes into
play).
nil means unset. false means explicitly set to a specific false value. empty
string means empty string. Clean programming languages like lua are very
careful to distinguish between those. rpm macro language is not careful, that’s
why the first google hits on rpm conditionals/comparisons is "it does not work
as you expect to", why no one writes complex logic rpm macro language, and why
redhat-rpm-config maintainers suspect any new macro submission to contain
dangerous quirks.
So get down your high horse and please implement things cleanly. And if you
insist, please do implement clean true and false keywords in the rpm macro
language to clean up the legacy mess there too.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-706048624___
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
Oh, looking closer: > @nim-nim wrote > [...] > exist because nil is not the same as false or empty Yes, nil is not same as empty. It however very explicitly is same as false, quoting https://www.lua.org/manual/5.4/manual.html#2.1: > Both nil and false make a condition false; they are collectively called false > values. Any other value makes a condition true. It doesn't get much more clearer than that. If you've assumed otherwise then that's your headache. If you look closer, this PR specifically *relies* on the fundamental language behavior of empty meaning true and nil being false. I'm not aware of anything in rpm assuming otherwise, but if there was that'd simply be a bug that needs fixing. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705566832___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
And none of us here would have the slightest clue as to what those fragments are supposed to be doing and in what context. As I'm sure you know perfectly well, the devil is in the details. If you're not interested in testing and reviewing *this* implementation that's fine, just don't come complaining about it later then. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705524090___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
I assume option without argument would be true while option with argument would be string with its value? -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705395114___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
https://src.fedoraproject.org/rpms/redhat-rpm-config/blob/master/f/common.lua#_30 exists because nil is not the same as false lua-side. And no I don’t remember out of hand all the problems it fixed. I definitely *do* *not* want to remember them. Debugging the corner cases induced by the approximations in existing rpm lua APIs were eating all my time macro dev before I took care to wrap it all in something solid and correct by default. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705493670___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
@nim-nim, if there's a concrete problem with how testing for option presence is handled here, lets hear it out. To me it seems to work exactly like one would expect it to, but sorry ponies not included. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705485732___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
@pmatilai given lua was added as a macro language because the built-in constructs were driving everyone crazy as soon as you needed some non-trivial processing I don’t find this argument too convincing :) Anyway lua has built-in conventions for booleans and nil values, you can’t dump blindly things as they exist non-lua side and expect things to work well lua-side. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705482062___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
The Lua native arguments have to match those of non-Lua macros, otherwise everybody goes crazy. For more advanced use there will be that option to let the macro do its own processing. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705460333___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
@hroncok from the lua language POW you need a tristate for flags: nil (not passed), true (passed as true), false (passed as false). Otherwise it’s real easy to be confused when handling things that have a default state absent any flag, or an antonym (with/without, quiet/verbose, etc), or One may argue that that does not matter given the primitive state of the current rpm argument parser, but what that actually tells me building a new API, that bakes existing parsing problems even deeper, is quite unwise. @pmatilai has the right general idea but it would probably be safer to expose arguments using an existing cli parsing API (such as lua argparse) and not embark in a journey to rediscover problems others mastered a long time ago. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705452828___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
...and done, now including actual documentation: https://github.com/rpm-software-management/rpm/pull/1383/commits/a3c336434f0abcfb5c17ef4718429f86873f7f4d @hroncok - basically yes, options without value are saved as empty strings in the Lua table so both boolean and value access work as you'd expect. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705411963___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
(This all is making me think if we couldn't actually arrange parametric macros to appear as native callable functions in Lua...) -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705310315___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
We can't pass NULL's in middle of ARGV_t, but that gave me an idea...
What we could do is pass pass options and a table and the arguments as another,
and prepend this to the anonymous Lua chunk from C side:
```
opt = select(1, ...)
arg = select(2, ...)
```
Which I think makes it all really nice on the Lua side:
```
%foo(a:b) %{lua
if opt.b
print('do b')
if opt.a
print('a value: '...opt.a)
for i in ipairs(i, arg) do
print(i, arg)
end
--- also these would work
if opt['b']:
print('do b')
print('first arg'..argv[1])
}
```
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-705309465___
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
I'd love to write:
```
%{lua: opt_z, arg = ...; }
```
i.e pass an argument for every option in the macro definition (using the order
specified in the definition). Use nil if the option was not used, 0/1 for
options with no argument.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-704832591___
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
https://lua.programmingpedia.net/en/tutorial/4475/variadic-arguments applies,
eg to walk through arguments:
> $ ./rpm --define "foo(..) %{lua: t = {...}; print('nargs '..rpm.expand('%{#}:
> ')); for i, v in ipairs(t) do print(i..':'..v..' '); end}" --eval "%foo 5 4 3
> 2 -z"
nargs 5: 1:5 2:4 3:3 4:2 5:-z
The question mark(s) revolve around what to do with options when present in
normal parametric macros. The current behavior is to pass them to Lua as
arguments, and %{#} macro carries the number of non-option arguments so you can
calculate how many option arguments there are.
> $ ./rpm --define "foo(z:) %{lua: t = {...}; print('nargs '..rpm.expand('%{#}:
> ')); for i, v in ipairs(t) do print(i..':'..v..' '); end}" --eval "%foo 5 4 3
> 2 -z7"
nargs 4: 1:-z7 2:5 3:4 4:3 5:2
Another possibility would be not passing options as arguments to Lua except
when custom option processing is selected, which would *kinda* make sense, but
also seems limiting.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-704290557___
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
Do you have an example of usage? It's hard to find for me in the tests. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-704199944___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
Implemented now as a part of #1383.
There are various ways this could be done, what's implemented now is that %{**}
macro, ie the unprocessed arguments to the macro, are passed to Lua as
arguments. The processed per-option macros are still accessible via
rpm.expand() of course. Feedback welcome.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-704171945___
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
I was about to explore this as well, thanks for filing this. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-628473072___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
#546 and #547 need to be considered when doing this. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1092#issuecomment-601710019___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
[Rpm-maint] [rpm-software-management/rpm] RFE: pass parametric macro options and arguments to Lua natively (#1092)
Accessing macro arguments via rpm.expand("%1") etc is tedious, non-intuitive
and subject to all sorts of expansion issues. Make macro options and arguments
available to Lua via native data structures (tables in practise), as if they
were just function arguments. Which they largely are.
PR #1063 was an experiment at doing this but relies on global structures which
is not a feasible approach, parametric macros can nest so we need to handle the
arguments via Lua stack. This requires enhancing the internal rpmlua wrapper
API around Lua.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/1092___
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint
