Feature preview
When multiple sites share a content model, how to prevent fields from interfering with each other? Limit the common aspects to the structure layer, and leave the differences in the value layer: the model and field names should be consistent across sites, default values and options should be configured separately for each site, and the keyword library should also be maintained separately for each site. Before making changes, confirm which templates and interfaces refer to the fields, and then verify the rendering results separately for each site to avoid affecting multiple brand pages at the same time.
How do you prevent fields from interfering with each other when using a shared content model across multiple sites? The principle is to only place the shared aspects at the structural level, with differences left at the value level: the model and field names are consistent across all sites, default values, option sets, and display text are configured separately for each site, and the keyword library is maintained separately for each site. AnQiCMS's multi-site management supports maintaining multiple independent sites on a single backend, suitable for multi-brand and multi-theme scenarios. The customizable content model can flexibly define the content structure, and when used together, it is necessary to define clear boundaries.
A content model commonly used items are divided into three categories: field names and types, field and template value relationships, and business significance of fields. These three categories are suitable for unification because they determine whether interfaces and templates can be reused. Items not suitable for sharing are default values, optional values, display text, and validation rules, which are the landing points of brand and regional differences.
Misplacing boundaries can lead to two symptoms: one is that fields with the same name have different meanings on different sites, the interface output is normal but the content is confused; the other is that a separate model is built for each site, the template cannot be reused, and the maintenance volume grows linearly with the number of sites.
It is recommended to name according to 'business object plus attribute', without site prefix or color and other visual information. The practice of putting the site name in the field prefix can distinguish in the short term, but in the long run, it will cause the same template to take different field names in two sites.
Differences should be handled by the configuration layer instead of new fields:
The keyword library centrally manages SEO keywords, and under multiple sites, they should be maintained separately by site. Centralized management solves the problem of having an editing entry in one backend, but does not mean that different brands can share the same batch of words. The consequence of sharing is that the automatic rules will fall the keywords of the main site onto the content of the sub-sites, causing the anchor text and description to be misaligned.
The same approach also applies to columns and tags: the naming is unified, and specific entries are maintained according to the site. Content that needs to be reused across sites should be carried by independent documents, not maintained by copying and pasting two copies.
The verification of field changes should be traced back by reference relationship. First confirm which templates use this field and which interfaces output it, then decide on the change method: renaming or deleting is a destructive change, which requires updating the templates and interface call parties at the same time; adding a new field is a compatible change, but it is necessary to check the performance of the default value on old content.
The station-level operation is also constrained at the interface layer. The MCP interface sets the exposure range according to the intent domain, and capabilities such as backup, upgrade, and multi-site are turned off by default (DefaultOff), and need to be explicitly enabled. Cross-site changes will not affect all sites simultaneously due to a single accidental call.
It is recommended to follow a fixed sequence of four steps: first, make changes in a single site and verify the rendering and interface output; then repeat on the second site to confirm that all differences are in the configuration layer; next, check if the path for batch import and interface writing is normal under the new field structure; finally, record the changes in the change list, indicating the affected sites and rollback methods.
Question: When the number of sites is large, should the model be split?Look at the difference ratio. If the field set has a high degree of overlap, continue to use a common set and leave the differences in the configuration; when the structural difference of a site is obvious, it is better to build a separate model for it than to insert conditional branches in the common model, which is more maintainable.
What are the safe practices for renaming fields?Add the target field, cut the template and interface references to the new field, and then disable the old field after checking for errors. Do this in three steps to avoid interrupting rendering and interfaces with one-time changes.
Should the column names be unified across multiple sites?It is recommended to unify the structure name, allowing different display names. Unifying names allows templates and interfaces to be reused, while display names are left for brand differentiation.
Preview
It is important to back up before opening the template editor, not after making changes. AnQiCMS's AI visual template editor can directly modify template code, and the larger the changes, the more necessary it is to have a rollback snapshot; the backup covers data and static files, starting from v3.6.5, it also expanded the template backup table, use it to store an original template first, then let the AI take over.
Preview
Whether the abstract is exposed depends on which field is taken from the list page, not the visible range written on the title. AnQiCMS uses user management and VIP grouping to set access permissions for different user groups, and the permissions need to be received at three places: list data acquisition, detail page, and interface return. Hiding only the detail page means that the abstract is still exposed in the list, and the first few paragraphs of the text can still be read.
Preview
Anchor text automatic linking added too much can really mess up the structure of the internal links, the chaos comes from one word pointing to multiple things, repeated stacking on the same page, and conflicts with the replacement of the entire station, rather than the function itself. The criterion for judgment is to see if the target pointed to by the same word is single and if there is an upper limit to the number of times it appears on each page. AnQiCMS manages anchor text according to the keyword library to do automatic internal linking, and is配合 the batch rewriting of the replacement of the entire station, the two have different responsibilities and cannot be mixed.