Comments on: v.sh part 1: wrapper script

By: Adventures in the forest of V « Fixpoint

Adventures in the forest of V « Fixpoint — Tue, 31 Mar 2020 19:08:30 +0000

[...] a thing would still require Perl, itself not an insignificant liability. While work had been underway to replace that, the results fell short of completeness, and from the ensuing discussion I decided [...]

By: JFW plan, week of 23 Mar. 2020 « Young Hands Club

JFW plan, week of 23 Mar. 2020 « Young Hands Club — Mon, 23 Mar 2020 07:34:47 +0000

[...] in scheduling errands first thing in the morning. [↩]I'm seeing something like bvt's v.sh as the way forward but the tried and mostly-true perl one as a present necessity. [...]

By: TMSR OS, February 2020 Statement « Dorion Mode

TMSR OS, February 2020 Statement « Dorion Mode — Wed, 04 Mar 2020 07:18:47 +0000

[...] his awk + posix shell + Ada Vtron in a 3 part series (1, 2, [...]

By: Mircea Popescu

Mircea Popescu — Wed, 05 Feb 2020 21:55:19 +0000

Alrighty!

By: bvt

bvt — Tue, 04 Feb 2020 07:53:55 +0000

For some reason I was expecting to see a hasher in there as well, vkeccak or w/e :p

Myeah, I was thinking about merging all of the tools into one, perhaps later. Re vkeccak, vpatch checks input and the output hashes, so there is no need for extra verification. And even in this case, ksum would do.

Really, why not ? I must specifically curl myself, why ? Can't it just have a

v mirror -- copy the given path locally.

such that v mirror "http://ossasepia.com/reference-code-shelf/" results in all the .vpatch and .sig files encountered dumped in a directory named for the domain ?

I am not against this, but I have never personally seen the need for myself; I can add this later.

I'm not much of a fan either. If memory serves it's some lulz introduced back when overoptimistic development attempted to make "a linux windows" (= ubuntu) on the flimsy (and meanwhile thoroughly disabused) hope that such would result into a linuxification of the windowstards, rather than a windowtardation of linux.

IIRC original Unix team wanted to prevent ls from outputting "." (this directory) and ".." (parent directory), but made a bug - did not check if there are other symbols after the initial dots, and later this behavior got standardized. The windowstardation must be the extended attributes added for Samba/CIFS.

But all that aside : no mention of manifest file ?

The manifest file is there, but was nothing to write about, so I dropped it in the writeup.

By: Mircea Popescu

Mircea Popescu — Tue, 04 Feb 2020 05:07:12 +0000

> A set of applications in Ada, one for extracting information out of vpatches (vfilter), calculation of dependencies among vpatches (vflow), for topological sorting based on the previously calculated dependency information (vtoposort), and for simplication of the dependency graph (vsimplify).

For some reason I was expecting to see a hasher in there as well, vkeccak or w/e :p

> No automatic initialization/syncing with a remote code shelf though.

Really, why not ? I must specifically curl myself, why ? Can't it just have a

v mirror -- copy the given path locally.

such that v mirror "http://ossasepia.com/reference-code-shelf/" results in all the .vpatch and .sig files encountered dumped in a directory named for the domain ?

> I do not really like "hidden" files, though

I'm not much of a fan either. If memory serves it's some lulz introduced back when overoptimistic development attempted to make "a linux windows" (= ubuntu) on the flimsy (and meanwhile thoroughly disabused) hope that such would result into a linuxification of the windowstards, rather than a windowtardation of linux.

But all that aside : no mention of manifest file ?

By: Stanislav Datskovskiy

Stanislav Datskovskiy — Mon, 03 Feb 2020 00:10:11 +0000

@bvt: Got it. (And, similarly, my PGP-eater also requires GNAT to be present on the machine (to have built Peh.))

By: bvt

bvt — Sun, 02 Feb 2020 21:13:32 +0000

@Diana
Thank you, good to hear that I'm not alone in this preference. The next post should come in ~3 days.

@Stanislav
Thanks. Re awk problem: I have never seriously considered awk for the core of the application -- I would have rather used shell+Ada if this was feasible; awk works great if the implemented application becomes short enough and data filtering is the task, but for numeric arrays Ada is much better. But in the context of vfilter and internall - awk anything but chokes, I was pleasantly surprised by it's performance.

By: Stanislav Datskovskiy

Stanislav Datskovskiy — Sun, 02 Feb 2020 17:26:29 +0000

This is very neat. I have a yet-unfinished "vtron in posix utils" to go with my PGP sig eater in posix utils, but it is -- just as you predicted -- limited in the max weight of vpatch that it can eat.

Seems to me that the simpler solution to "awk chokes on N GB vpatches" would be to fix awk though, rather than write a new util? Normally this would be impractical, but if you're baking a new linux kit, why not fix awk ?

By: Diana Coman

Diana Coman — Sun, 02 Feb 2020 12:53:19 +0000

I am expecting that people will ask to change "wot" to ".wot"

No, I really prefer wot (and seals) too! I don't really see any specific reason to have those as hidden directories anyway and I'm not aware of any rationale or discussion on this previously either.

At a first read, it looks quite good to me. Looking forward to the rest of the series too.