Saturday, 2023-05-27

*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC05:04
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc05:16
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC06:13
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc06:14
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC08:21
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc08:22
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC08:29
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc08:38
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC09:17
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc09:18
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC09:37
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc09:38
ghostmansd[m]lkcl, I just recalled that we still haven't discussed if another task which causes less conflicts and debates is available. :-)09:58
ghostmansd[m]BTW, searching in IRC logs yields 404:
lkcli know i haven't added the cgi-script11:34
lkclthe model simulator is a really high business priority and i do not have the time to take on that task as well11:35
lkcli am dealing with the specification which is extremely complex and taking all of my energy and focus11:35
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC11:36
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc11:36
lkcli know what is happening with the model: it's not yet at the point where iterative feedback is possible, so there is "nothing to understand, no context, nothing"11:37
lkclonce it produces actual output based on input then understanding will follow and it takes off from there11:38
ghostmansd[m]I need some task which has some space for manoeuvre and free will.12:04
ghostmansd[m]Considering the overall feedback, I really don't want to handle this task, _especially_ if this is high priority.12:06
lkclthere's two high-priority tasks: i am tackling the spec feedback and communication with the OPF ISA WG12:07
lkcli cannot do both12:07
lkclthe model is a very simple task12:07
lkclfar simpler than imagined12:07
lkclthere are no external imports - not one single one - bar absolute basic imports and certainly absolutely nothing from any of our repos.12:08
lkclits only input is an ASCII file in a well-defined format12:08
lkclits only output is an ASCII file in a markdown-table format12:08
ghostmansd[m]"a very simple task" — not if any solution goes under heavy criticism and is later totally refactored according to personal preferences.12:09
ghostmansd[m]If this is a high-priority task, then you should either relax the requirements or pass it to someone else. I also have a personal situation here and go through quite complicated stuff right now, and I don't want to waste my time and energy on debates and looking at the code which was shitty according to someone else's standards. It's fine per se to have different opinions, but if somebody writes a totally different code than you expect than perhap12:14
ghostmansd[m]the wrong person to implement the functionality is chosen.12:14
lkclthey're not "personal preferences", there are specific reasons for keeping it short and brutally simple12:17
lkcli may have to actually present this code directly to customers12:17
lkclas part of a presentation12:17
lkclif it has complex classes that span multiple pages, i cannot put those onto a slide12:17
lkclif it has concepts that are not easy to read by a non-expert python programmer, i cannot explain them in a reasonable amount of time12:18
lkclso the code has to be written in a style that an (intelligent) beginner of only 6-months programming experience can easily grasp12:20
ghostmansd[m]Why should you post code? You should post the principle and the algorithm, not implementation details.12:20
lkclbecause this is very much like a "config file"12:20
lkclwhere the pipeline lengths will be parameters that can be changed by the customer, to experiment12:21
lkclit also means plenty of explanatory code-comments are needed12:21
lkclit's a very different style from one where an advanced programmer would create classes, dataclasses, use regex's because they cope with complex data patterns and so on12:22
lkclok i'm done with shift-and-add
lkcli'm still doing a spec-change from the new LD/ST-Indexed format12:54
lkclghostmansd[m], actually, could you do that?12:55
ghostmansd[m]Update the spec? What's the repo, libreriscv?13:05
ghostmansd[m]Also binutils must be updated too. :-)13:05
ghostmansd[m]Hm, I suspect there already is SH field, isn't it?13:06
ghostmansd[m]If these occupy different bits, this might cause it to be renamed in binutils.13:06
ghostmansd[m]Paul suggested SHM113:08
lkclyes, i... sigh13:10
lkclyes that's a real simple one to look at13:11
lkclcan you do it under #1091?13:11
lkclit's something-to-do13:11
lkclthe LD-ST-Indexed thing is a little more involved, it's literally introducing a new EXTRA format that is a *mixed* EXTRA3-EXTRA3-EXTRA2 form13:12
lkclby using bits 6&7 as MASK_SRC13:13
lkclthat involves quite a shuffle in (esp of the base classes)13:14
ghostmansd[m]I'll take a look at fields renaming, => SHM1, including binutils, np14:12
ghostmansd[m]And also 1079, later14:13
lkcldon't make it SHM1 - that's a misunderstanding by the IBM team.  the name was suggested "Shift Minus 1" which is *not correct*15:21
lkclSH1 would be fine because actually it really is SH+115:21
lkclSHP1 would do but i am not keen on 4-character-long fields15:21
lkclya know what? screw it: SHP1 is pretty clear15:52
lkclsigh you'll have to follow the action-log here to do the same thing, sigh
ghostmansd[m]np, I'll handle it :-)16:33
ghostmansd[m]Already checked the logs to realize that I have to repeat what's done already :-)16:34
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC19:04
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc19:06
*** markos_ <markos_!~markos_@user/markos/x-1838887> has quit IRC19:37
*** markos_ <markos_!> has joined #libre-soc19:51
*** markos_ <markos_!~markos_@user/markos/x-1838887> has joined #libre-soc19:51
lkclghostmansd[m], well sigh if the name is better it has to be done, wark-wark20:29

Generated by 2.17.1 by Marius Gedminas - find it at!