----- Forwarded message from Luis Mochán <moc...@fis.unam.mx> -----
Date: Sun, 08 Dec 2024 20:17:06 -0600
From: Luis Mochán <moc...@fis.unam.mx>
To: "Ed ." <ej...@hotmail.com>
Subject: Re: approx fails for complex types
User-Agent: K-9 Mail for Android
Hi Ed,
I realized that afterwards, it is a common risk of macro expansions. I
was in some hurry and didn't realize the error in the pr until
now. Thanks for the advice.
Luis
On December 8, 2024 4:14:14 PM CST, "Ed ." <ej...@hotmail.com> wrote:
>Hi Luis,
>
>There's nothing wrong with the expansion of the macro, but there was something
>wrong with the code I wrote, which was for real types:
>
>value = a * b*b; /* correct */
>
>but for complex values:
>
>value = a * b*b+c*c; /* wrong, I meant a * (b*b+c*c) */
>
>Thank you for the pull request (#507)! I adjusted it to add your new test to
>the existing tests in t/primitive-misc.t rather than adding a new file. I've
>also changed it so that the brackets are added only for the complex case.
>
>Notes to all:
>
> *
>if you do add a new file, you need to add it to MANIFEST as well. That was why
>the PR failed the release tests, which check that everything is either in
>MANIFEST or is skipped in MANIFEST.SKIP. You can run those yourself locally
>with RELEASE_TESTING=1 prove -l xt
> *
>if you do make a PR, please check it does in fact pass the CI, which due to
>the above this didn't.
> *
>when making a PR, please make a new local branch rather than using your master
>branch - as it is, you'll need to `git reset --hard HEAD^^` to erase your
>commits before doing a `git pull --rebase` to get my updated ones
>
>Above all those points though, please do make PRs! It's hugely valuable and
>makes PDL better for everyone. It's only that bearing the above points in mind
>will make my life a bit easier :-)
>
>I've merged the adjusted PR, and it will be in the next PDL. That shouldn't be
>too much longer, as I've nearly finished separating out the
>externally-dependent modules as discussed elsewhere.
>
>Best regards,
>Ed
>
>________________________________
>From: Luis Mochan <moc...@icf.unam.mx>
>Sent: 08 December 2024 8:03 PM
>To: Ed . <ej...@hotmail.com>
>Cc: pdl-devel <pdl-devel@lists.sourceforge.net>; perldl
><pdl-gene...@lists.sourceforge.net>
>Subject: approx fails for complex types
>
>Hi All,
>
>I found a mysterious error when comparing complex types. For instance,
>
> pdl> p approx(i,5*i)
> 1
> pdl> p approx(5*i,i)
> 0
>
>i.e., i is close to 5i (wrong) but 5i is not close to i (correct).
>
>I found a (provisional?) fix, by simply using parenthesis around
> PDL_IF_GENTYPE_REAL(...)
>in Primitive.pd, but I guess the macro would have to be checked and I
>don't know if it might have produced other errors elsewhere.
>
>Best regards,
>Luis
>
>
>--
>
> o
>W. Luis Mochán, | tel:(52)(777)329-1734 /<(*)
>Instituto de Ciencias Físicas, UNAM | fax:(52)(777)317-5388 `>/ /\
>Av. Universidad s/n CP 62210 | (*)/\/ \
>Cuernavaca, Morelos, México | moc...@fis.unam.mx /\_/\__/
>GPG: 791EB9EB, C949 3F81 6D9B 1191 9A16 C2DF 5F0A C52B 791E B9EB
--
Enviado desde mi celular; disculpe la horrografía
----- End forwarded message -----
--
o
W. Luis Mochán, | tel:(52)(777)329-1734 /<(*)
Instituto de Ciencias Físicas, UNAM | fax:(52)(777)317-5388 `>/ /\
Av. Universidad s/n CP 62210 | (*)/\/ \
Cuernavaca, Morelos, México | moc...@fis.unam.mx /\_/\__/
GPG: 791EB9EB, C949 3F81 6D9B 1191 9A16 C2DF 5F0A C52B 791E B9EB
_______________________________________________
pdl-devel mailing list
pdl-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/pdl-devel
