Re: lynx-dev PATCH [dev21]: source caching, take 3

[Leonid Pauzner]

12-Apr-99 19:08 address@hidden wrote:
>> Grmbl... with HTDisplayPartial() defined private in HTFormat.c, I had to
>> move a fair amount of code there from GridText.c; so I'm resubmitting
>> the patch as a replacement again.  I've checked this one, and at least
>> on this end, patch -p0 had no problems.

> patch got confused, I think, due to the similar chunks interwoven with
> Vlad's patch.  I merged a copy by hand

Think we should wait a little until dev22 will be out
since several people are out of sync while submitting patches these days
over and over.

One more problem with SOURCE_CACHE code found:
when you are toggling to "\" source mode
and then _immediately_ go to any internal page ("=" Info or "<-" History, etc.)
that next page got in source mode also. Mistake.

(well, I comment out the check for "http" and do my testing locally:
actually I run O)ptions menu, press "\" and then press "=").
Seems HTOutputFormat will not restored after SOURCE_CHACHE operations,
I mention one thing: LYUCPopAssumed() used in HTLoad() cycle
but obviously lost in Cache_Thrue... code. No chartrans problems, though.
More discussion after dev22 out.


> --- ./src/LYMainLoop.c.orig   Tue Mar 30 12:10:37 1999
> +++ ./src/LYMainLoop.c        Sun Apr 11 20:58:43 1999
> @@ -2049,6 +2069,12 @@
>               LYUCPushAssumed(HTMainAnchor);
>               HTOutputFormat = WWW_SOURCE;
>           }
> +#ifdef SOURCE_CACHE
> +         if (HTreparse_document()) {
> +             refresh_screen = TRUE;
> +             break;
> +         }
> +#endif
>           LYforce_no_cache = TRUE;
>           FREE(curdoc.address); /* so it doesn't get pushed */
>           break;
> @@ -2125,11 +2151,13 @@
>                                  0, 0) == FALSE) {
>               HTInfoMsg(WILL_NOT_RELOAD_DOC);
>           } else {
> +#ifndef SOURCE_CACHE
>               HTuncache_current_document();
>               StrAllocCopy(newdoc.address, curdoc.address);
>               FREE(curdoc.address);
>               newdoc.line = curdoc.line;
>               newdoc.link = curdoc.link;
> +#endif
>           }
>           if (historical_comments)
>               historical_comments = FALSE;